Merge branch 'main' into gl/chore/use-base-model-for-llms

docs/en/concepts/llms.mdx

@@ -1212,7 +1212,7 @@ Learn how to get the most out of your LLM configuration:
     ```python
 import httpx
 from crewai import LLM
-from crewai.llms.hooks import BaseInterceptor
+from crewai.llm.hooks import BaseInterceptor
 
 class CustomInterceptor(BaseInterceptor[httpx.Request, httpx.Response]):
     """Custom interceptor to modify requests and responses."""

lib/crewai/src/crewai/__init__.py

@@ -8,8 +8,8 @@ from crewai.crew import Crew
 from crewai.crews.crew_output import CrewOutput
 from crewai.flow.flow import Flow
 from crewai.knowledge.knowledge import Knowledge
-from crewai.llm import LLM
-from crewai.llms.base_llm import BaseLLM
+from crewai.llm.base_llm import BaseLLM
+from crewai.llm.core import LLM
 from crewai.process import Process
 from crewai.task import Task
 from crewai.tasks.llm_guardrail import LLMGuardrail

lib/crewai/src/crewai/agent/core.py

@@ -39,7 +39,7 @@ from crewai.knowledge.knowledge import Knowledge
 from crewai.knowledge.source.base_knowledge_source import BaseKnowledgeSource
 from crewai.knowledge.utils.knowledge_utils import extract_knowledge_context
 from crewai.lite_agent import LiteAgent
-from crewai.llms.base_llm import BaseLLM
+from crewai.llm.base_llm import BaseLLM
 from crewai.mcp import (
     MCPClient,
     MCPServerConfig,
@@ -633,7 +633,7 @@ class Agent(BaseAgent):
             )
 
         self.agent_executor = CrewAgentExecutor(
-            llm=self.llm,
+            llm=self.llm,  # type: ignore[arg-type]
             task=task,  # type: ignore[arg-type]
             agent=self,
             crew=self.crew,
@@ -810,6 +810,7 @@ class Agent(BaseAgent):
         from crewai.tools.base_tool import BaseTool
         from crewai.tools.mcp_native_tool import MCPNativeTool
 
+        transport: StdioTransport | HTTPTransport | SSETransport
         if isinstance(mcp_config, MCPServerStdio):
             transport = StdioTransport(
                 command=mcp_config.command,
@@ -903,10 +904,12 @@ class Agent(BaseAgent):
                                 server_name=server_name,
                                 run_context=None,
                             )
-                            if mcp_config.tool_filter(context, tool):
+                            # Try new signature first
+                            if mcp_config.tool_filter(context, tool):  # type: ignore[arg-type,call-arg]
                                 filtered_tools.append(tool)
                         except (TypeError, AttributeError):
-                            if mcp_config.tool_filter(tool):
+                            # Fallback to old signature
+                            if mcp_config.tool_filter(tool):  # type: ignore[arg-type,call-arg]
                                 filtered_tools.append(tool)
                     else:
                         # Not callable - include tool
@@ -981,7 +984,9 @@ class Agent(BaseAgent):
         path = parsed.path.replace("/", "_").strip("_")
         return f"{domain}_{path}" if path else domain
 
-    def _get_mcp_tool_schemas(self, server_params: dict) -> dict[str, dict]:
+    def _get_mcp_tool_schemas(
+        self, server_params: dict[str, Any]
+    ) -> dict[str, dict[str, Any]]:
         """Get tool schemas from MCP server for wrapper creation with caching."""
         server_url = server_params["url"]
 
@@ -995,7 +1000,7 @@ class Agent(BaseAgent):
                 self._logger.log(
                     "debug", f"Using cached MCP tool schemas for {server_url}"
                 )
-                return cached_data
+                return cast(dict[str, dict[str, Any]], cached_data)
 
         try:
             schemas = asyncio.run(self._get_mcp_tool_schemas_async(server_params))
@@ -1013,7 +1018,7 @@ class Agent(BaseAgent):
 
     async def _get_mcp_tool_schemas_async(
         self, server_params: dict[str, Any]
-    ) -> dict[str, dict]:
+    ) -> dict[str, dict[str, Any]]:
         """Async implementation of MCP tool schema retrieval with timeouts and retries."""
         server_url = server_params["url"]
         return await self._retry_mcp_discovery(
@@ -1021,7 +1026,7 @@ class Agent(BaseAgent):
         )
 
     async def _retry_mcp_discovery(
-        self, operation_func, server_url: str
+        self, operation_func: Any, server_url: str
     ) -> dict[str, dict[str, Any]]:
         """Retry MCP discovery operation with exponential backoff, avoiding try-except in loop."""
         last_error = None
@@ -1052,7 +1057,7 @@ class Agent(BaseAgent):
 
     @staticmethod
     async def _attempt_mcp_discovery(
-        operation_func, server_url: str
+        operation_func: Any, server_url: str
     ) -> tuple[dict[str, dict[str, Any]] | None, str, bool]:
         """Attempt single MCP discovery operation and return (result, error_message, should_retry)."""
         try:
@@ -1156,13 +1161,13 @@ class Agent(BaseAgent):
                     Field(..., description=field_description),
                 )
             else:
-                field_definitions[field_name] = (
+                field_definitions[field_name] = (  # type: ignore[assignment]
                     field_type | None,
                     Field(default=None, description=field_description),
                 )
 
         model_name = f"{tool_name.replace('-', '_').replace(' ', '_')}Schema"
-        return create_model(model_name, **field_definitions)
+        return create_model(model_name, **field_definitions)  # type: ignore[no-any-return,call-overload]
 
     def _json_type_to_python(self, field_schema: dict[str, Any]) -> type:
         """Convert JSON Schema type to Python type.
@@ -1182,16 +1187,16 @@ class Agent(BaseAgent):
                 if "const" in option:
                     types.append(str)
                 else:
-                    types.append(self._json_type_to_python(option))
+                    types.append(self._json_type_to_python(option))  # type: ignore[arg-type]
             unique_types = list(set(types))
             if len(unique_types) > 1:
                 result = unique_types[0]
                 for t in unique_types[1:]:
-                    result = result | t
+                    result = result | t  # type: ignore[assignment]
                 return result
             return unique_types[0]
 
-        type_mapping = {
+        type_mapping: dict[str, type] = {
             "string": str,
             "number": float,
             "integer": int,
@@ -1200,10 +1205,10 @@ class Agent(BaseAgent):
             "object": dict,
         }
 
-        return type_mapping.get(json_type, Any)
+        return type_mapping.get(json_type or "", Any)
 
     @staticmethod
-    def _fetch_amp_mcp_servers(mcp_name: str) -> list[dict]:
+    def _fetch_amp_mcp_servers(mcp_name: str) -> list[dict[str, Any]]:
         """Fetch MCP server configurations from CrewAI AMP API."""
         # TODO: Implement AMP API call to "integrations/mcps" endpoint
         # Should return list of server configs with URLs

lib/crewai/src/crewai/agents/agent_builder/base_agent.py

@@ -137,7 +137,7 @@ class BaseAgent(BaseModel, ABC, metaclass=AgentMeta):
         default=False,
         description="Enable agent to delegate and ask questions among each other.",
     )
-    tools: list[BaseTool] | None = Field(
+    tools: list[BaseTool] = Field(
         default_factory=list, description="Tools at agents' disposal"
     )
     max_iter: int = Field(
@@ -161,7 +161,7 @@ class BaseAgent(BaseModel, ABC, metaclass=AgentMeta):
         description="An instance of the ToolsHandler class.",
     )
     tools_results: list[dict[str, Any]] = Field(
-        default=[], description="Results of the tools used by the agent."
+        default_factory=list, description="Results of the tools used by the agent."
     )
     max_tokens: int | None = Field(
         default=None, description="Maximum number of tokens for the agent's execution."
@@ -265,7 +265,7 @@ class BaseAgent(BaseModel, ABC, metaclass=AgentMeta):
         if not mcps:
             return mcps
 
-        validated_mcps = []
+        validated_mcps: list[str | MCPServerConfig] = []
         for mcp in mcps:
             if isinstance(mcp, str):
                 if mcp.startswith(("https://", "crewai-amp:")):

lib/crewai/src/crewai/agents/crew_agent_executor.py

@@ -51,7 +51,7 @@ if TYPE_CHECKING:
     from crewai.agent import Agent
     from crewai.agents.tools_handler import ToolsHandler
     from crewai.crew import Crew
-    from crewai.llms.base_llm import BaseLLM
+    from crewai.llm.base_llm import BaseLLM
     from crewai.task import Task
     from crewai.tools.base_tool import BaseTool
     from crewai.tools.structured_tool import CrewStructuredTool

lib/crewai/src/crewai/cli/crew_chat.py

@@ -14,7 +14,8 @@ import tomli
 from crewai.cli.utils import read_toml
 from crewai.cli.version import get_crewai_version
 from crewai.crew import Crew
-from crewai.llm import LLM, BaseLLM
+from crewai.llm import LLM
+from crewai.llm.base_llm import BaseLLM
 from crewai.types.crew_chat import ChatInputField, ChatInputs
 from crewai.utilities.llm_utils import create_llm
 from crewai.utilities.printer import Printer

lib/crewai/src/crewai/crew.py

@@ -56,8 +56,8 @@ from crewai.events.types.crew_events import (
 from crewai.flow.flow_trackable import FlowTrackable
 from crewai.knowledge.knowledge import Knowledge
 from crewai.knowledge.source.base_knowledge_source import BaseKnowledgeSource
-from crewai.llm import LLM
-from crewai.llms.base_llm import BaseLLM
+from crewai.llm.base_llm import BaseLLM
+from crewai.llm.core import LLM
 from crewai.memory.entity.entity_memory import EntityMemory
 from crewai.memory.external.external_memory import ExternalMemory
 from crewai.memory.long_term.long_term_memory import LongTermMemory

lib/crewai/src/crewai/events/event_listener.py

@@ -89,7 +89,7 @@ from crewai.events.types.tool_usage_events import (
     ToolUsageStartedEvent,
 )
 from crewai.events.utils.console_formatter import ConsoleFormatter
-from crewai.llm import LLM
+from crewai.llm.core import LLM
 from crewai.task import Task
 from crewai.telemetry.telemetry import Telemetry
 from crewai.utilities import Logger

lib/crewai/src/crewai/experimental/evaluation/base_evaluator.py

@@ -7,7 +7,7 @@ from pydantic import BaseModel, Field
 
 from crewai.agent import Agent
 from crewai.agents.agent_builder.base_agent import BaseAgent
-from crewai.llm import BaseLLM
+from crewai.llm.base_llm import BaseLLM
 from crewai.task import Task
 from crewai.utilities.llm_utils import create_llm
 

lib/crewai/src/crewai/lite_agent.py

@@ -39,8 +39,8 @@ from crewai.events.types.agent_events import (
 from crewai.events.types.logging_events import AgentLogsExecutionEvent
 from crewai.flow.flow_trackable import FlowTrackable
 from crewai.lite_agent_output import LiteAgentOutput
-from crewai.llm import LLM
-from crewai.llms.base_llm import BaseLLM
+from crewai.llm.base_llm import BaseLLM
+from crewai.llm.core import LLM
 from crewai.tools.base_tool import BaseTool
 from crewai.tools.structured_tool import CrewStructuredTool
 from crewai.utilities.agent_utils import (
@@ -504,7 +504,7 @@ class LiteAgent(FlowTrackable, BaseModel):
             AgentFinish: The final result of the agent execution.
         """
         # Execute the agent loop
-        formatted_answer = None
+        formatted_answer: AgentAction | AgentFinish | None = None
         while not isinstance(formatted_answer, AgentFinish):
             try:
                 if has_reached_max_iterations(self._iterations, self.max_iterations):
@@ -553,7 +553,8 @@ class LiteAgent(FlowTrackable, BaseModel):
                         show_logs=self._show_logs,
                     )
 
-                self._append_message(formatted_answer.text, role="assistant")
+                if formatted_answer is not None:
+                    self._append_message(formatted_answer.text, role="assistant")
             except OutputParserError as e:  # noqa: PERF203
                 self._printer.print(
                     content="Failed to parse LLM output. Retrying...",

lib/crewai/src/crewai/llm/__init__.py

@@ -0,0 +1,4 @@
+from crewai.llm.core import LLM
+
+
+__all__ = ["LLM"]

lib/crewai/src/crewai/llm/base_llm.py

@@ -0,0 +1,588 @@
+"""Base LLM abstract class for CrewAI.
+
+This module provides the abstract base class for all LLM implementations
+in CrewAI, including common functionality for native SDK implementations.
+"""
+
+from __future__ import annotations
+
+from abc import ABC, abstractmethod
+from datetime import datetime
+import json
+import logging
+import os
+import re
+from typing import TYPE_CHECKING, Any, Final
+
+from dotenv import load_dotenv
+import httpx
+from pydantic import BaseModel, Field, field_validator
+
+from crewai.events.event_bus import crewai_event_bus
+from crewai.events.types.llm_events import (
+    LLMCallCompletedEvent,
+    LLMCallFailedEvent,
+    LLMCallStartedEvent,
+    LLMCallType,
+    LLMStreamChunkEvent,
+)
+from crewai.events.types.tool_usage_events import (
+    ToolUsageErrorEvent,
+    ToolUsageFinishedEvent,
+    ToolUsageStartedEvent,
+)
+from crewai.llm.hooks.base import BaseInterceptor
+from crewai.llm.internal.meta import LLMMeta
+from crewai.types.usage_metrics import UsageMetrics
+
+
+if TYPE_CHECKING:
+    from crewai.agent.core import Agent
+    from crewai.task import Task
+    from crewai.tools.base_tool import BaseTool
+    from crewai.utilities.types import LLMMessage
+
+
+load_dotenv()
+
+DEFAULT_CONTEXT_WINDOW_SIZE: Final[int] = 4096
+DEFAULT_SUPPORTS_STOP_WORDS: Final[bool] = True
+_JSON_EXTRACTION_PATTERN: Final[re.Pattern[str]] = re.compile(r"\{.*}", re.DOTALL)
+
+
+class BaseLLM(BaseModel, ABC, metaclass=LLMMeta):
+    """Abstract base class for LLM implementations.
+
+    This class defines the interface that all LLM implementations must follow.
+    Users can extend this class to create custom LLM implementations that don't
+    rely on litellm's authentication mechanism.
+
+    Custom LLM implementations should handle error cases gracefully, including
+    timeouts, authentication failures, and malformed responses. They should also
+    implement proper validation for input parameters and provide clear error
+    messages when things go wrong.
+
+    Attributes:
+        model: The model identifier/name.
+        temperature: Optional temperature setting for response generation.
+        stop: A list of stop sequences that the LLM should use to stop generation.
+    """
+
+    # Core fields
+    model: str = Field(..., description="The model identifier/name")
+    temperature: float | None = Field(
+        None, description="Temperature setting for response generation"
+    )
+    api_key: str | None = Field(None, description="API key for authentication")
+    base_url: str | None = Field(None, description="Base URL for API requests")
+    provider: str = Field(
+        default="openai", description="Provider name (openai, anthropic, etc.)"
+    )
+    stop: list[str] = Field(
+        default_factory=list,
+        description="Stop sequences for generation",
+        alias="stop_sequences",
+    )
+
+    # Internal fields
+    is_litellm: bool = Field(
+        default=False, description="Whether this instance uses LiteLLM"
+    )
+    interceptor: BaseInterceptor[httpx.Request, httpx.Response] | None = Field(
+        default=None, description="HTTP request/response interceptor"
+    )
+    _token_usage: dict[str, int] = {
+        "total_tokens": 0,
+        "prompt_tokens": 0,
+        "completion_tokens": 0,
+        "successful_requests": 0,
+        "cached_prompt_tokens": 0,
+    }
+
+    @field_validator("api_key", mode="after")
+    @classmethod
+    def _validate_api_key(cls, value: str | None) -> str | None:
+        """Validate API key for authentication.
+
+        Args:
+            value: API key value or None
+
+        Returns:
+            API key from environment if not provided, or the original value
+        """
+        if value is None:
+            cls_name = cls.__name__
+            provider_prefix = cls_name.replace("Completion", "").upper()
+            env_var = f"{provider_prefix}_API_KEY"
+            value = os.getenv(env_var)
+        return value
+
+    @field_validator("stop", mode="before")
+    @classmethod
+    def _normalize_stop(cls, value: Any) -> list[str]:
+        """Normalize stop sequences to a list.
+
+        Args:
+            value: Stop sequences as string, list, or None
+
+        Returns:
+            Normalized list of stop sequences
+        """
+        if value is None:
+            return []
+        if isinstance(value, str):
+            return [value]
+        if isinstance(value, list):
+            return value
+        return []
+
+    @property
+    def additional_params(self) -> dict[str, Any]:
+        """Get additional parameters stored as extra fields.
+
+        Returns:
+            Dictionary of additional parameters
+        """
+        return self.__pydantic_extra__ or {}
+
+    @additional_params.setter
+    def additional_params(self, value: dict[str, Any]) -> None:
+        """Set additional parameters as extra fields.
+
+        Args:
+            value: Dictionary of additional parameters to set
+        """
+        if not isinstance(value, dict):
+            raise ValueError("additional_params must be a dictionary")
+        if self.__pydantic_extra__ is None:
+            self.__pydantic_extra__ = {}
+        self.__pydantic_extra__.update(value)
+
+    @abstractmethod
+    def call(
+        self,
+        messages: str | list[LLMMessage],
+        tools: list[dict[str, BaseTool]] | None = None,
+        callbacks: list[Any] | None = None,
+        available_functions: dict[str, Any] | None = None,
+        from_task: Task | None = None,
+        from_agent: Agent | None = None,
+        response_model: type[BaseModel] | None = None,
+    ) -> str | Any:
+        """Call the LLM with the given messages.
+
+        Args:
+            messages: Input messages for the LLM.
+                     Can be a string or list of message dictionaries.
+                     If string, it will be converted to a single user message.
+                     If list, each dict must have 'role' and 'content' keys.
+            tools: Optional list of tool schemas for function calling.
+                  Each tool should define its name, description, and parameters.
+            callbacks: Optional list of callback functions to be executed
+                      during and after the LLM call.
+            available_functions: Optional dict mapping function names to callables
+                               that can be invoked by the LLM.
+            from_task: Optional task caller to be used for the LLM call.
+            from_agent: Optional agent caller to be used for the LLM call.
+            response_model: Optional response model to be used for the LLM call.
+
+        Returns:
+            Either a text response from the LLM (str) or
+            the result of a tool function call (Any).
+
+        Raises:
+            ValueError: If the messages format is invalid.
+            TimeoutError: If the LLM request times out.
+            RuntimeError: If the LLM request fails for other reasons.
+        """
+
+    def _convert_tools_for_interference(
+        self, tools: list[dict[str, BaseTool]]
+    ) -> list[dict[str, BaseTool]]:
+        """Convert tools to a format that can be used for interference.
+
+        Args:
+            tools: List of tools to convert.
+
+        Returns:
+            List of converted tools (default implementation returns as-is)
+        """
+        return tools
+
+    def supports_stop_words(self) -> bool:
+        """Check if the LLM supports stop words.
+
+        Returns:
+            True if the LLM supports stop words, False otherwise.
+        """
+        return DEFAULT_SUPPORTS_STOP_WORDS
+
+    def _supports_stop_words_implementation(self) -> bool:
+        """Check if stop words are configured for this LLM instance.
+
+        Native providers can override supports_stop_words() to return this value
+        to ensure consistent behavior based on whether stop words are actually configured.
+
+        Returns:
+            True if stop words are configured and can be applied
+        """
+        return bool(self.stop)
+
+    def _apply_stop_words(self, content: str) -> str:
+        """Apply stop words to truncate response content.
+
+        This method provides consistent stop word behavior across all native SDK providers.
+        Native providers should call this method to post-process their responses.
+
+        Args:
+            content: The raw response content from the LLM
+
+        Returns:
+            Content truncated at the first occurrence of any stop word
+
+        Example:
+            >>> llm = MyNativeLLM(stop=["Observation:", "Final Answer:"])
+            >>> response = (
+            ...     "I need to search.\\n\\nAction: search\\nObservation: Found results"
+            ... )
+            >>> llm._apply_stop_words(response)
+            "I need to search.\\n\\nAction: search"
+        """
+        if not self.stop or not content:
+            return content
+
+        # Find the earliest occurrence of any stop word
+        earliest_stop_pos = len(content)
+        found_stop_word = None
+
+        for stop_word in self.stop:
+            stop_pos = content.find(stop_word)
+            if stop_pos != -1 and stop_pos < earliest_stop_pos:
+                earliest_stop_pos = stop_pos
+                found_stop_word = stop_word
+
+        # Truncate at the stop word if found
+        if found_stop_word is not None:
+            truncated = content[:earliest_stop_pos].strip()
+            logging.debug(
+                f"Applied stop word '{found_stop_word}' at position {earliest_stop_pos}"
+            )
+            return truncated
+
+        return content
+
+    def get_context_window_size(self) -> int:
+        """Get the context window size for the LLM.
+
+        Returns:
+            The number of tokens/characters the model can handle.
+        """
+        # Default implementation - subclasses should override with model-specific values
+        return DEFAULT_CONTEXT_WINDOW_SIZE
+
+    # Common helper methods for native SDK implementations
+
+    def _emit_call_started_event(
+        self,
+        messages: str | list[LLMMessage],
+        tools: list[dict[str, BaseTool]] | None = None,
+        callbacks: list[Any] | None = None,
+        available_functions: dict[str, Any] | None = None,
+        from_task: Task | None = None,
+        from_agent: Agent | None = None,
+    ) -> None:
+        """Emit LLM call started event."""
+        if not hasattr(crewai_event_bus, "emit"):
+            raise ValueError("crewai_event_bus does not have an emit method") from None
+
+        crewai_event_bus.emit(
+            self,
+            event=LLMCallStartedEvent(
+                messages=messages,
+                tools=tools,
+                callbacks=callbacks,
+                available_functions=available_functions,
+                from_task=from_task,
+                from_agent=from_agent,
+                model=self.model,
+            ),
+        )
+
+    def _emit_call_completed_event(
+        self,
+        response: Any,
+        call_type: LLMCallType,
+        from_task: Task | None = None,
+        from_agent: Agent | None = None,
+        messages: str | list[dict[str, Any]] | None = None,
+    ) -> None:
+        """Emit LLM call completed event."""
+        crewai_event_bus.emit(
+            self,
+            event=LLMCallCompletedEvent(
+                messages=messages,
+                response=response,
+                call_type=call_type,
+                from_task=from_task,
+                from_agent=from_agent,
+                model=self.model,
+            ),
+        )
+
+    def _emit_call_failed_event(
+        self,
+        error: str,
+        from_task: Task | None = None,
+        from_agent: Agent | None = None,
+    ) -> None:
+        """Emit LLM call failed event."""
+        if not hasattr(crewai_event_bus, "emit"):
+            raise ValueError("crewai_event_bus does not have an emit method") from None
+
+        crewai_event_bus.emit(
+            self,
+            event=LLMCallFailedEvent(
+                error=error,
+                from_task=from_task,
+                from_agent=from_agent,
+            ),
+        )
+
+    def _emit_stream_chunk_event(
+        self,
+        chunk: str,
+        from_task: Task | None = None,
+        from_agent: Agent | None = None,
+        tool_call: dict[str, Any] | None = None,
+    ) -> None:
+        """Emit stream chunk event."""
+        if not hasattr(crewai_event_bus, "emit"):
+            raise ValueError("crewai_event_bus does not have an emit method") from None
+
+        crewai_event_bus.emit(
+            self,
+            event=LLMStreamChunkEvent(
+                chunk=chunk,
+                tool_call=tool_call,
+                from_task=from_task,
+                from_agent=from_agent,
+            ),
+        )
+
+    def _handle_tool_execution(
+        self,
+        function_name: str,
+        function_args: dict[str, Any],
+        available_functions: dict[str, Any],
+        from_task: Task | None = None,
+        from_agent: Agent | None = None,
+    ) -> str | None:
+        """Handle tool execution with proper event emission.
+
+        Args:
+            function_name: Name of the function to execute
+            function_args: Arguments to pass to the function
+            available_functions: Dict of available functions
+            from_task: Optional task object
+            from_agent: Optional agent object
+
+        Returns:
+            Result of function execution or None if function not found
+        """
+        if function_name not in available_functions:
+            logging.warning(
+                f"Function '{function_name}' not found in available functions"
+            )
+            return None
+
+        try:
+            # Emit tool usage started event
+            started_at = datetime.now()
+
+            crewai_event_bus.emit(
+                self,
+                event=ToolUsageStartedEvent(
+                    tool_name=function_name,
+                    tool_args=function_args,
+                    from_agent=from_agent,
+                    from_task=from_task,
+                ),
+            )
+
+            # Execute the function
+            fn = available_functions[function_name]
+            result = fn(**function_args)
+
+            # Emit tool usage finished event
+            crewai_event_bus.emit(
+                self,
+                event=ToolUsageFinishedEvent(
+                    output=result,
+                    tool_name=function_name,
+                    tool_args=function_args,
+                    started_at=started_at,
+                    finished_at=datetime.now(),
+                    from_task=from_task,
+                    from_agent=from_agent,
+                ),
+            )
+
+            # Emit LLM call completed event for tool call
+            self._emit_call_completed_event(
+                response=result,
+                call_type=LLMCallType.TOOL_CALL,
+                from_task=from_task,
+                from_agent=from_agent,
+            )
+
+            return str(result)
+
+        except Exception as e:
+            error_msg = f"Error executing function '{function_name}': {e!s}"
+            logging.error(error_msg)
+
+            # Emit tool usage error event
+            if not hasattr(crewai_event_bus, "emit"):
+                raise ValueError(
+                    "crewai_event_bus does not have an emit method"
+                ) from None
+
+            crewai_event_bus.emit(
+                self,
+                event=ToolUsageErrorEvent(
+                    tool_name=function_name,
+                    tool_args=function_args,
+                    error=error_msg,
+                    from_task=from_task,
+                    from_agent=from_agent,
+                ),
+            )
+
+            # Emit LLM call failed event
+            self._emit_call_failed_event(
+                error=error_msg,
+                from_task=from_task,
+                from_agent=from_agent,
+            )
+
+            return None
+
+    def _format_messages(self, messages: str | list[LLMMessage]) -> list[LLMMessage]:
+        """Convert messages to standard format.
+
+        Args:
+            messages: Input messages (string or list of message dicts)
+
+        Returns:
+            List of message dictionaries with 'role' and 'content' keys
+
+        Raises:
+            ValueError: If message format is invalid
+        """
+        if isinstance(messages, str):
+            return [{"role": "user", "content": messages}]
+
+        # Validate message format
+        for i, msg in enumerate(messages):
+            if not isinstance(msg, dict):
+                raise ValueError(f"Message at index {i} must be a dictionary")
+            if "role" not in msg or "content" not in msg:
+                raise ValueError(
+                    f"Message at index {i} must have 'role' and 'content' keys"
+                )
+
+        return messages
+
+    @staticmethod
+    def _validate_structured_output(
+        response: str,
+        response_format: type[BaseModel] | None,
+    ) -> str | BaseModel:
+        """Validate and parse structured output.
+
+        Args:
+            response: Raw response string
+            response_format: Optional Pydantic model for structured output
+
+        Returns:
+            Parsed response (BaseModel instance if response_format provided, otherwise string)
+
+        Raises:
+            ValueError: If structured output validation fails
+        """
+        if response_format is None:
+            return response
+
+        try:
+            # Try to parse as JSON first
+            if response.strip().startswith("{") or response.strip().startswith("["):
+                data = json.loads(response)
+                return response_format.model_validate(data)
+
+            json_match = _JSON_EXTRACTION_PATTERN.search(response)
+            if json_match:
+                data = json.loads(json_match.group())
+                return response_format.model_validate(data)
+
+            raise ValueError("No JSON found in response")
+
+        except (json.JSONDecodeError, ValueError) as e:
+            logging.warning(f"Failed to parse structured output: {e}")
+            raise ValueError(
+                f"Failed to parse response into {response_format.__name__}: {e}"
+            ) from e
+
+    @staticmethod
+    def _extract_provider(model: str) -> str:
+        """Extract provider from model string.
+
+        Args:
+            model: Model string (e.g., 'openai/gpt-4' or 'gpt-4')
+
+        Returns:
+            Provider name (e.g., 'openai')
+        """
+        if "/" in model:
+            return model.partition("/")[0]
+        return "openai"  # Default provider
+
+    def _track_token_usage_internal(self, usage_data: dict[str, Any]) -> None:
+        """Track token usage internally in the LLM instance.
+
+        Args:
+            usage_data: Token usage data from the API response
+        """
+        # Extract tokens in a provider-agnostic way
+        prompt_tokens = (
+            usage_data.get("prompt_tokens")
+            or usage_data.get("prompt_token_count")
+            or usage_data.get("input_tokens")
+            or 0
+        )
+
+        completion_tokens = (
+            usage_data.get("completion_tokens")
+            or usage_data.get("candidates_token_count")
+            or usage_data.get("output_tokens")
+            or 0
+        )
+
+        cached_tokens = (
+            usage_data.get("cached_tokens")
+            or usage_data.get("cached_prompt_tokens")
+            or 0
+        )
+
+        self._token_usage["prompt_tokens"] += prompt_tokens
+        self._token_usage["completion_tokens"] += completion_tokens
+        self._token_usage["total_tokens"] += prompt_tokens + completion_tokens
+        self._token_usage["successful_requests"] += 1
+        self._token_usage["cached_prompt_tokens"] += cached_tokens
+
+    def get_token_usage_summary(self) -> UsageMetrics:
+        """Get summary of token usage for this LLM instance.
+
+        Returns:
+            Dictionary with token usage totals
+        """
+        return UsageMetrics(**self._token_usage)

lib/crewai/src/crewai/llm/constants.py

@@ -0,0 +1,587 @@
+from typing import Literal, TypeAlias
+
+
+SupportedNativeProviders: TypeAlias = Literal[
+    "openai",
+    "anthropic",
+    "claude",
+    "azure",
+    "azure_openai",
+    "google",
+    "gemini",
+    "bedrock",
+    "aws",
+]
+
+SUPPORTED_NATIVE_PROVIDERS: list[SupportedNativeProviders] = [
+    "openai",
+    "anthropic",
+    "claude",
+    "azure",
+    "azure_openai",
+    "google",
+    "gemini",
+    "bedrock",
+    "aws",
+]
+
+
+OpenAIModels: TypeAlias = Literal[
+    "gpt-3.5-turbo",
+    "gpt-3.5-turbo-0125",
+    "gpt-3.5-turbo-0301",
+    "gpt-3.5-turbo-0613",
+    "gpt-3.5-turbo-1106",
+    "gpt-3.5-turbo-16k",
+    "gpt-3.5-turbo-16k-0613",
+    "gpt-3.5-turbo-instruct",
+    "gpt-3.5-turbo-instruct-0914",
+    "gpt-4",
+    "gpt-4-0125-preview",
+    "gpt-4-0314",
+    "gpt-4-0613",
+    "gpt-4-1106-preview",
+    "gpt-4-32k",
+    "gpt-4-32k-0314",
+    "gpt-4-32k-0613",
+    "gpt-4-turbo",
+    "gpt-4-turbo-2024-04-09",
+    "gpt-4-turbo-preview",
+    "gpt-4-vision-preview",
+    "gpt-4.1",
+    "gpt-4.1-2025-04-14",
+    "gpt-4.1-mini",
+    "gpt-4.1-mini-2025-04-14",
+    "gpt-4.1-nano",
+    "gpt-4.1-nano-2025-04-14",
+    "gpt-4o",
+    "gpt-4o-2024-05-13",
+    "gpt-4o-2024-08-06",
+    "gpt-4o-2024-11-20",
+    "gpt-4o-audio-preview",
+    "gpt-4o-audio-preview-2024-10-01",
+    "gpt-4o-audio-preview-2024-12-17",
+    "gpt-4o-audio-preview-2025-06-03",
+    "gpt-4o-mini",
+    "gpt-4o-mini-2024-07-18",
+    "gpt-4o-mini-audio-preview",
+    "gpt-4o-mini-audio-preview-2024-12-17",
+    "gpt-4o-mini-realtime-preview",
+    "gpt-4o-mini-realtime-preview-2024-12-17",
+    "gpt-4o-mini-search-preview",
+    "gpt-4o-mini-search-preview-2025-03-11",
+    "gpt-4o-mini-transcribe",
+    "gpt-4o-mini-tts",
+    "gpt-4o-realtime-preview",
+    "gpt-4o-realtime-preview-2024-10-01",
+    "gpt-4o-realtime-preview-2024-12-17",
+    "gpt-4o-realtime-preview-2025-06-03",
+    "gpt-4o-search-preview",
+    "gpt-4o-search-preview-2025-03-11",
+    "gpt-4o-transcribe",
+    "gpt-4o-transcribe-diarize",
+    "gpt-5",
+    "gpt-5-2025-08-07",
+    "gpt-5-chat",
+    "gpt-5-chat-latest",
+    "gpt-5-codex",
+    "gpt-5-mini",
+    "gpt-5-mini-2025-08-07",
+    "gpt-5-nano",
+    "gpt-5-nano-2025-08-07",
+    "gpt-5-pro",
+    "gpt-5-pro-2025-10-06",
+    "gpt-5-search-api",
+    "gpt-5-search-api-2025-10-14",
+    "gpt-audio",
+    "gpt-audio-2025-08-28",
+    "gpt-audio-mini",
+    "gpt-audio-mini-2025-10-06",
+    "gpt-image-1",
+    "gpt-image-1-mini",
+    "gpt-realtime",
+    "gpt-realtime-2025-08-28",
+    "gpt-realtime-mini",
+    "gpt-realtime-mini-2025-10-06",
+    "o1",
+    "o1-preview",
+    "o1-2024-12-17",
+    "o1-mini",
+    "o1-mini-2024-09-12",
+    "o1-pro",
+    "o1-pro-2025-03-19",
+    "o3-mini",
+    "o3",
+    "o4-mini",
+    "whisper-1",
+]
+OPENAI_MODELS: list[OpenAIModels] = [
+    "gpt-3.5-turbo",
+    "gpt-3.5-turbo-0125",
+    "gpt-3.5-turbo-0301",
+    "gpt-3.5-turbo-0613",
+    "gpt-3.5-turbo-1106",
+    "gpt-3.5-turbo-16k",
+    "gpt-3.5-turbo-16k-0613",
+    "gpt-3.5-turbo-instruct",
+    "gpt-3.5-turbo-instruct-0914",
+    "gpt-4",
+    "gpt-4-0125-preview",
+    "gpt-4-0314",
+    "gpt-4-0613",
+    "gpt-4-1106-preview",
+    "gpt-4-32k",
+    "gpt-4-32k-0314",
+    "gpt-4-32k-0613",
+    "gpt-4-turbo",
+    "gpt-4-turbo-2024-04-09",
+    "gpt-4-turbo-preview",
+    "gpt-4-vision-preview",
+    "gpt-4.1",
+    "gpt-4.1-2025-04-14",
+    "gpt-4.1-mini",
+    "gpt-4.1-mini-2025-04-14",
+    "gpt-4.1-nano",
+    "gpt-4.1-nano-2025-04-14",
+    "gpt-4o",
+    "gpt-4o-2024-05-13",
+    "gpt-4o-2024-08-06",
+    "gpt-4o-2024-11-20",
+    "gpt-4o-audio-preview",
+    "gpt-4o-audio-preview-2024-10-01",
+    "gpt-4o-audio-preview-2024-12-17",
+    "gpt-4o-audio-preview-2025-06-03",
+    "gpt-4o-mini",
+    "gpt-4o-mini-2024-07-18",
+    "gpt-4o-mini-audio-preview",
+    "gpt-4o-mini-audio-preview-2024-12-17",
+    "gpt-4o-mini-realtime-preview",
+    "gpt-4o-mini-realtime-preview-2024-12-17",
+    "gpt-4o-mini-search-preview",
+    "gpt-4o-mini-search-preview-2025-03-11",
+    "gpt-4o-mini-transcribe",
+    "gpt-4o-mini-tts",
+    "gpt-4o-realtime-preview",
+    "gpt-4o-realtime-preview-2024-10-01",
+    "gpt-4o-realtime-preview-2024-12-17",
+    "gpt-4o-realtime-preview-2025-06-03",
+    "gpt-4o-search-preview",
+    "gpt-4o-search-preview-2025-03-11",
+    "gpt-4o-transcribe",
+    "gpt-4o-transcribe-diarize",
+    "gpt-5",
+    "gpt-5-2025-08-07",
+    "gpt-5-chat",
+    "gpt-5-chat-latest",
+    "gpt-5-codex",
+    "gpt-5-mini",
+    "gpt-5-mini-2025-08-07",
+    "gpt-5-nano",
+    "gpt-5-nano-2025-08-07",
+    "gpt-5-pro",
+    "gpt-5-pro-2025-10-06",
+    "gpt-5-search-api",
+    "gpt-5-search-api-2025-10-14",
+    "gpt-audio",
+    "gpt-audio-2025-08-28",
+    "gpt-audio-mini",
+    "gpt-audio-mini-2025-10-06",
+    "gpt-image-1",
+    "gpt-image-1-mini",
+    "gpt-realtime",
+    "gpt-realtime-2025-08-28",
+    "gpt-realtime-mini",
+    "gpt-realtime-mini-2025-10-06",
+    "o1",
+    "o1-preview",
+    "o1-2024-12-17",
+    "o1-mini",
+    "o1-mini-2024-09-12",
+    "o1-pro",
+    "o1-pro-2025-03-19",
+    "o3-mini",
+    "o3",
+    "o4-mini",
+    "whisper-1",
+]
+
+
+AnthropicModels: TypeAlias = Literal[
+    "claude-3-7-sonnet-latest",
+    "claude-3-7-sonnet-20250219",
+    "claude-3-5-haiku-latest",
+    "claude-3-5-haiku-20241022",
+    "claude-haiku-4-5",
+    "claude-haiku-4-5-20251001",
+    "claude-sonnet-4-20250514",
+    "claude-sonnet-4-0",
+    "claude-4-sonnet-20250514",
+    "claude-sonnet-4-5",
+    "claude-sonnet-4-5-20250929",
+    "claude-3-5-sonnet-latest",
+    "claude-3-5-sonnet-20241022",
+    "claude-3-5-sonnet-20240620",
+    "claude-opus-4-0",
+    "claude-opus-4-20250514",
+    "claude-4-opus-20250514",
+    "claude-opus-4-1",
+    "claude-opus-4-1-20250805",
+    "claude-3-opus-latest",
+    "claude-3-opus-20240229",
+    "claude-3-sonnet-20240229",
+    "claude-3-haiku-latest",
+    "claude-3-haiku-20240307",
+]
+ANTHROPIC_MODELS: list[AnthropicModels] = [
+    "claude-3-7-sonnet-latest",
+    "claude-3-7-sonnet-20250219",
+    "claude-3-5-haiku-latest",
+    "claude-3-5-haiku-20241022",
+    "claude-haiku-4-5",
+    "claude-haiku-4-5-20251001",
+    "claude-sonnet-4-20250514",
+    "claude-sonnet-4-0",
+    "claude-4-sonnet-20250514",
+    "claude-sonnet-4-5",
+    "claude-sonnet-4-5-20250929",
+    "claude-3-5-sonnet-latest",
+    "claude-3-5-sonnet-20241022",
+    "claude-3-5-sonnet-20240620",
+    "claude-opus-4-0",
+    "claude-opus-4-20250514",
+    "claude-4-opus-20250514",
+    "claude-opus-4-1",
+    "claude-opus-4-1-20250805",
+    "claude-3-opus-latest",
+    "claude-3-opus-20240229",
+    "claude-3-sonnet-20240229",
+    "claude-3-haiku-latest",
+    "claude-3-haiku-20240307",
+]
+
+GeminiModels: TypeAlias = Literal[
+    "gemini-2.5-pro",
+    "gemini-2.5-pro-preview-03-25",
+    "gemini-2.5-pro-preview-05-06",
+    "gemini-2.5-pro-preview-06-05",
+    "gemini-2.5-flash",
+    "gemini-2.5-flash-preview-05-20",
+    "gemini-2.5-flash-preview-04-17",
+    "gemini-2.5-flash-image",
+    "gemini-2.5-flash-image-preview",
+    "gemini-2.5-flash-lite",
+    "gemini-2.5-flash-lite-preview-06-17",
+    "gemini-2.5-flash-preview-09-2025",
+    "gemini-2.5-flash-lite-preview-09-2025",
+    "gemini-2.5-flash-preview-tts",
+    "gemini-2.5-pro-preview-tts",
+    "gemini-2.5-computer-use-preview-10-2025",
+    "gemini-2.0-flash",
+    "gemini-2.0-flash-001",
+    "gemini-2.0-flash-exp",
+    "gemini-2.0-flash-exp-image-generation",
+    "gemini-2.0-flash-lite",
+    "gemini-2.0-flash-lite-001",
+    "gemini-2.0-flash-lite-preview",
+    "gemini-2.0-flash-lite-preview-02-05",
+    "gemini-2.0-flash-preview-image-generation",
+    "gemini-2.0-flash-thinking-exp",
+    "gemini-2.0-flash-thinking-exp-01-21",
+    "gemini-2.0-flash-thinking-exp-1219",
+    "gemini-2.0-pro-exp",
+    "gemini-2.0-pro-exp-02-05",
+    "gemini-exp-1206",
+    "gemini-1.5-pro",
+    "gemini-1.5-flash",
+    "gemini-1.5-flash-8b",
+    "gemini-flash-latest",
+    "gemini-flash-lite-latest",
+    "gemini-pro-latest",
+    "gemini-2.0-flash-live-001",
+    "gemini-live-2.5-flash-preview",
+    "gemini-2.5-flash-live-preview",
+    "gemini-robotics-er-1.5-preview",
+    "gemini-gemma-2-27b-it",
+    "gemini-gemma-2-9b-it",
+    "gemma-3-1b-it",
+    "gemma-3-4b-it",
+    "gemma-3-12b-it",
+    "gemma-3-27b-it",
+    "gemma-3n-e2b-it",
+    "gemma-3n-e4b-it",
+    "learnlm-2.0-flash-experimental",
+]
+GEMINI_MODELS: list[GeminiModels] = [
+    "gemini-2.5-pro",
+    "gemini-2.5-pro-preview-03-25",
+    "gemini-2.5-pro-preview-05-06",
+    "gemini-2.5-pro-preview-06-05",
+    "gemini-2.5-flash",
+    "gemini-2.5-flash-preview-05-20",
+    "gemini-2.5-flash-preview-04-17",
+    "gemini-2.5-flash-image",
+    "gemini-2.5-flash-image-preview",
+    "gemini-2.5-flash-lite",
+    "gemini-2.5-flash-lite-preview-06-17",
+    "gemini-2.5-flash-preview-09-2025",
+    "gemini-2.5-flash-lite-preview-09-2025",
+    "gemini-2.5-flash-preview-tts",
+    "gemini-2.5-pro-preview-tts",
+    "gemini-2.5-computer-use-preview-10-2025",
+    "gemini-2.0-flash",
+    "gemini-2.0-flash-001",
+    "gemini-2.0-flash-exp",
+    "gemini-2.0-flash-exp-image-generation",
+    "gemini-2.0-flash-lite",
+    "gemini-2.0-flash-lite-001",
+    "gemini-2.0-flash-lite-preview",
+    "gemini-2.0-flash-lite-preview-02-05",
+    "gemini-2.0-flash-preview-image-generation",
+    "gemini-2.0-flash-thinking-exp",
+    "gemini-2.0-flash-thinking-exp-01-21",
+    "gemini-2.0-flash-thinking-exp-1219",
+    "gemini-2.0-pro-exp",
+    "gemini-2.0-pro-exp-02-05",
+    "gemini-exp-1206",
+    "gemini-1.5-pro",
+    "gemini-1.5-flash",
+    "gemini-1.5-flash-8b",
+    "gemini-flash-latest",
+    "gemini-flash-lite-latest",
+    "gemini-pro-latest",
+    "gemini-2.0-flash-live-001",
+    "gemini-live-2.5-flash-preview",
+    "gemini-2.5-flash-live-preview",
+    "gemini-robotics-er-1.5-preview",
+    "gemini-gemma-2-27b-it",
+    "gemini-gemma-2-9b-it",
+    "gemma-3-1b-it",
+    "gemma-3-4b-it",
+    "gemma-3-12b-it",
+    "gemma-3-27b-it",
+    "gemma-3n-e2b-it",
+    "gemma-3n-e4b-it",
+    "learnlm-2.0-flash-experimental",
+]
+
+
+AzureModels: TypeAlias = Literal[
+    "gpt-3.5-turbo",
+    "gpt-3.5-turbo-0301",
+    "gpt-3.5-turbo-0613",
+    "gpt-3.5-turbo-16k",
+    "gpt-3.5-turbo-16k-0613",
+    "gpt-35-turbo",
+    "gpt-35-turbo-0125",
+    "gpt-35-turbo-1106",
+    "gpt-35-turbo-16k-0613",
+    "gpt-35-turbo-instruct-0914",
+    "gpt-4",
+    "gpt-4-0314",
+    "gpt-4-0613",
+    "gpt-4-1106-preview",
+    "gpt-4-0125-preview",
+    "gpt-4-32k",
+    "gpt-4-32k-0314",
+    "gpt-4-32k-0613",
+    "gpt-4-turbo",
+    "gpt-4-turbo-2024-04-09",
+    "gpt-4-vision",
+    "gpt-4o",
+    "gpt-4o-2024-05-13",
+    "gpt-4o-2024-08-06",
+    "gpt-4o-2024-11-20",
+    "gpt-4o-mini",
+    "gpt-5",
+    "o1",
+    "o1-mini",
+    "o1-preview",
+    "o3-mini",
+    "o3",
+    "o4-mini",
+]
+AZURE_MODELS: list[AzureModels] = [
+    "gpt-3.5-turbo",
+    "gpt-3.5-turbo-0301",
+    "gpt-3.5-turbo-0613",
+    "gpt-3.5-turbo-16k",
+    "gpt-3.5-turbo-16k-0613",
+    "gpt-35-turbo",
+    "gpt-35-turbo-0125",
+    "gpt-35-turbo-1106",
+    "gpt-35-turbo-16k-0613",
+    "gpt-35-turbo-instruct-0914",
+    "gpt-4",
+    "gpt-4-0314",
+    "gpt-4-0613",
+    "gpt-4-1106-preview",
+    "gpt-4-0125-preview",
+    "gpt-4-32k",
+    "gpt-4-32k-0314",
+    "gpt-4-32k-0613",
+    "gpt-4-turbo",
+    "gpt-4-turbo-2024-04-09",
+    "gpt-4-vision",
+    "gpt-4o",
+    "gpt-4o-2024-05-13",
+    "gpt-4o-2024-08-06",
+    "gpt-4o-2024-11-20",
+    "gpt-4o-mini",
+    "gpt-5",
+    "o1",
+    "o1-mini",
+    "o1-preview",
+    "o3-mini",
+    "o3",
+    "o4-mini",
+]
+
+
+BedrockModels: TypeAlias = Literal[
+    "ai21.jamba-1-5-large-v1:0",
+    "ai21.jamba-1-5-mini-v1:0",
+    "amazon.nova-lite-v1:0",
+    "amazon.nova-lite-v1:0:24k",
+    "amazon.nova-lite-v1:0:300k",
+    "amazon.nova-micro-v1:0",
+    "amazon.nova-micro-v1:0:128k",
+    "amazon.nova-micro-v1:0:24k",
+    "amazon.nova-premier-v1:0",
+    "amazon.nova-premier-v1:0:1000k",
+    "amazon.nova-premier-v1:0:20k",
+    "amazon.nova-premier-v1:0:8k",
+    "amazon.nova-premier-v1:0:mm",
+    "amazon.nova-pro-v1:0",
+    "amazon.nova-pro-v1:0:24k",
+    "amazon.nova-pro-v1:0:300k",
+    "amazon.titan-text-express-v1",
+    "amazon.titan-text-express-v1:0:8k",
+    "amazon.titan-text-lite-v1",
+    "amazon.titan-text-lite-v1:0:4k",
+    "amazon.titan-tg1-large",
+    "anthropic.claude-3-5-haiku-20241022-v1:0",
+    "anthropic.claude-3-5-sonnet-20240620-v1:0",
+    "anthropic.claude-3-5-sonnet-20241022-v2:0",
+    "anthropic.claude-3-7-sonnet-20250219-v1:0",
+    "anthropic.claude-3-haiku-20240307-v1:0",
+    "anthropic.claude-3-haiku-20240307-v1:0:200k",
+    "anthropic.claude-3-haiku-20240307-v1:0:48k",
+    "anthropic.claude-3-opus-20240229-v1:0",
+    "anthropic.claude-3-opus-20240229-v1:0:12k",
+    "anthropic.claude-3-opus-20240229-v1:0:200k",
+    "anthropic.claude-3-opus-20240229-v1:0:28k",
+    "anthropic.claude-3-sonnet-20240229-v1:0",
+    "anthropic.claude-3-sonnet-20240229-v1:0:200k",
+    "anthropic.claude-3-sonnet-20240229-v1:0:28k",
+    "anthropic.claude-haiku-4-5-20251001-v1:0",
+    "anthropic.claude-instant-v1:2:100k",
+    "anthropic.claude-opus-4-1-20250805-v1:0",
+    "anthropic.claude-opus-4-20250514-v1:0",
+    "anthropic.claude-sonnet-4-20250514-v1:0",
+    "anthropic.claude-sonnet-4-5-20250929-v1:0",
+    "anthropic.claude-v2:0:100k",
+    "anthropic.claude-v2:0:18k",
+    "anthropic.claude-v2:1:18k",
+    "anthropic.claude-v2:1:200k",
+    "cohere.command-r-plus-v1:0",
+    "cohere.command-r-v1:0",
+    "cohere.rerank-v3-5:0",
+    "deepseek.r1-v1:0",
+    "meta.llama3-1-70b-instruct-v1:0",
+    "meta.llama3-1-8b-instruct-v1:0",
+    "meta.llama3-2-11b-instruct-v1:0",
+    "meta.llama3-2-1b-instruct-v1:0",
+    "meta.llama3-2-3b-instruct-v1:0",
+    "meta.llama3-2-90b-instruct-v1:0",
+    "meta.llama3-3-70b-instruct-v1:0",
+    "meta.llama3-70b-instruct-v1:0",
+    "meta.llama3-8b-instruct-v1:0",
+    "meta.llama4-maverick-17b-instruct-v1:0",
+    "meta.llama4-scout-17b-instruct-v1:0",
+    "mistral.mistral-7b-instruct-v0:2",
+    "mistral.mistral-large-2402-v1:0",
+    "mistral.mistral-small-2402-v1:0",
+    "mistral.mixtral-8x7b-instruct-v0:1",
+    "mistral.pixtral-large-2502-v1:0",
+    "openai.gpt-oss-120b-1:0",
+    "openai.gpt-oss-20b-1:0",
+    "qwen.qwen3-32b-v1:0",
+    "qwen.qwen3-coder-30b-a3b-v1:0",
+    "twelvelabs.pegasus-1-2-v1:0",
+]
+BEDROCK_MODELS: list[BedrockModels] = [
+    "ai21.jamba-1-5-large-v1:0",
+    "ai21.jamba-1-5-mini-v1:0",
+    "amazon.nova-lite-v1:0",
+    "amazon.nova-lite-v1:0:24k",
+    "amazon.nova-lite-v1:0:300k",
+    "amazon.nova-micro-v1:0",
+    "amazon.nova-micro-v1:0:128k",
+    "amazon.nova-micro-v1:0:24k",
+    "amazon.nova-premier-v1:0",
+    "amazon.nova-premier-v1:0:1000k",
+    "amazon.nova-premier-v1:0:20k",
+    "amazon.nova-premier-v1:0:8k",
+    "amazon.nova-premier-v1:0:mm",
+    "amazon.nova-pro-v1:0",
+    "amazon.nova-pro-v1:0:24k",
+    "amazon.nova-pro-v1:0:300k",
+    "amazon.titan-text-express-v1",
+    "amazon.titan-text-express-v1:0:8k",
+    "amazon.titan-text-lite-v1",
+    "amazon.titan-text-lite-v1:0:4k",
+    "amazon.titan-tg1-large",
+    "anthropic.claude-3-5-haiku-20241022-v1:0",
+    "anthropic.claude-3-5-sonnet-20240620-v1:0",
+    "anthropic.claude-3-5-sonnet-20241022-v2:0",
+    "anthropic.claude-3-7-sonnet-20250219-v1:0",
+    "anthropic.claude-3-haiku-20240307-v1:0",
+    "anthropic.claude-3-haiku-20240307-v1:0:200k",
+    "anthropic.claude-3-haiku-20240307-v1:0:48k",
+    "anthropic.claude-3-opus-20240229-v1:0",
+    "anthropic.claude-3-opus-20240229-v1:0:12k",
+    "anthropic.claude-3-opus-20240229-v1:0:200k",
+    "anthropic.claude-3-opus-20240229-v1:0:28k",
+    "anthropic.claude-3-sonnet-20240229-v1:0",
+    "anthropic.claude-3-sonnet-20240229-v1:0:200k",
+    "anthropic.claude-3-sonnet-20240229-v1:0:28k",
+    "anthropic.claude-haiku-4-5-20251001-v1:0",
+    "anthropic.claude-instant-v1:2:100k",
+    "anthropic.claude-opus-4-1-20250805-v1:0",
+    "anthropic.claude-opus-4-20250514-v1:0",
+    "anthropic.claude-sonnet-4-20250514-v1:0",
+    "anthropic.claude-sonnet-4-5-20250929-v1:0",
+    "anthropic.claude-v2:0:100k",
+    "anthropic.claude-v2:0:18k",
+    "anthropic.claude-v2:1:18k",
+    "anthropic.claude-v2:1:200k",
+    "cohere.command-r-plus-v1:0",
+    "cohere.command-r-v1:0",
+    "cohere.rerank-v3-5:0",
+    "deepseek.r1-v1:0",
+    "meta.llama3-1-70b-instruct-v1:0",
+    "meta.llama3-1-8b-instruct-v1:0",
+    "meta.llama3-2-11b-instruct-v1:0",
+    "meta.llama3-2-1b-instruct-v1:0",
+    "meta.llama3-2-3b-instruct-v1:0",
+    "meta.llama3-2-90b-instruct-v1:0",
+    "meta.llama3-3-70b-instruct-v1:0",
+    "meta.llama3-70b-instruct-v1:0",
+    "meta.llama3-8b-instruct-v1:0",
+    "meta.llama4-maverick-17b-instruct-v1:0",
+    "meta.llama4-scout-17b-instruct-v1:0",
+    "mistral.mistral-7b-instruct-v0:2",
+    "mistral.mistral-large-2402-v1:0",
+    "mistral.mistral-small-2402-v1:0",
+    "mistral.mixtral-8x7b-instruct-v0:1",
+    "mistral.pixtral-large-2502-v1:0",
+    "openai.gpt-oss-120b-1:0",
+    "openai.gpt-oss-20b-1:0",
+    "qwen.qwen3-32b-v1:0",
+    "qwen.qwen3-coder-30b-a3b-v1:0",
+    "twelvelabs.pegasus-1-2-v1:0",
+]
+
+SupportedModels: TypeAlias = (
+    OpenAIModels | AnthropicModels | GeminiModels | AzureModels | BedrockModels
+)

lib/crewai/src/crewai/llm/core.py

@@ -20,9 +20,7 @@ from typing import (
 )
 
 from dotenv import load_dotenv
-import httpx
 from pydantic import BaseModel, Field
-from typing_extensions import Self
 
 from crewai.events.event_bus import crewai_event_bus
 from crewai.events.types.llm_events import (
@@ -37,14 +35,7 @@ from crewai.events.types.tool_usage_events import (
     ToolUsageFinishedEvent,
     ToolUsageStartedEvent,
 )
-from crewai.llms.base_llm import BaseLLM
-from crewai.llms.constants import (
-    ANTHROPIC_MODELS,
-    AZURE_MODELS,
-    BEDROCK_MODELS,
-    GEMINI_MODELS,
-    OPENAI_MODELS,
-)
+from crewai.llm.base_llm import BaseLLM
 from crewai.utilities import InternalInstructor
 from crewai.utilities.exceptions.context_window_exceeding_exception import (
     LLMContextLengthExceededError,
@@ -61,7 +52,6 @@ if TYPE_CHECKING:
     from litellm.utils import supports_response_schema
 
     from crewai.agent.core import Agent
-    from crewai.llms.hooks.base import BaseInterceptor
     from crewai.task import Task
     from crewai.tools.base_tool import BaseTool
     from crewai.utilities.types import LLMMessage
@@ -327,249 +317,57 @@ class AccumulatedToolArgs(BaseModel):
 
 
 class LLM(BaseLLM):
-    completion_cost: float | None = None
+    """LiteLLM-based LLM implementation for CrewAI.
 
-    def __new__(cls, model: str, is_litellm: bool = False, **kwargs: Any) -> LLM:
-        """Factory method that routes to native SDK or falls back to LiteLLM.
+    This class provides LiteLLM integration for models not covered by native providers.
+    The metaclass (LLMMeta) automatically routes to native providers when appropriate.
+    """
 
-        Routing priority:
-            1. If 'provider' kwarg is present, use that provider with constants
-            2. If only 'model' kwarg, use constants to infer provider
-            3. If "/" in model name:
-               - Check if prefix is a native provider (openai/anthropic/azure/bedrock/gemini)
-               - If yes, validate model against constants
-               - If valid, route to native SDK; otherwise route to LiteLLM
-        """
-        if not model or not isinstance(model, str):
-            raise ValueError("Model must be a non-empty string")
+    # LiteLLM-specific fields
+    completion_cost: float | None = Field(None, description="Cost of completion")
+    timeout: float | int | None = Field(None, description="Request timeout")
+    top_p: float | None = Field(None, description="Top-p sampling parameter")
+    n: int | None = Field(None, description="Number of completions to generate")
+    max_completion_tokens: int | None = Field(
+        None, description="Maximum completion tokens"
+    )
+    max_tokens: int | float | None = Field(None, description="Maximum total tokens")
+    presence_penalty: float | None = Field(None, description="Presence penalty")
+    frequency_penalty: float | None = Field(None, description="Frequency penalty")
+    logit_bias: dict[int, float] | None = Field(None, description="Logit bias")
+    response_format: type[BaseModel] | None = Field(
+        None, description="Response format model"
+    )
+    seed: int | None = Field(None, description="Random seed for reproducibility")
+    logprobs: int | None = Field(None, description="Log probabilities to return")
+    top_logprobs: int | None = Field(None, description="Top log probabilities")
+    api_base: str | None = Field(None, description="API base URL (alias for base_url)")
+    api_version: str | None = Field(None, description="API version")
+    callbacks: list[Any] | None = Field(None, description="Callback functions")
+    context_window_size: int = Field(0, description="Context window size in tokens")
+    reasoning_effort: Literal["none", "low", "medium", "high"] | None = Field(
+        None, description="Reasoning effort level"
+    )
+    is_anthropic: bool = Field(False, description="Whether model is from Anthropic")
+    stream: bool = Field(False, description="Whether to stream responses")
 
-        explicit_provider = kwargs.get("provider")
-
-        if explicit_provider:
-            provider = explicit_provider
-            use_native = True
-            model_string = model
-        elif "/" in model:
-            prefix, _, model_part = model.partition("/")
-
-            provider_mapping = {
-                "openai": "openai",
-                "anthropic": "anthropic",
-                "claude": "anthropic",
-                "azure": "azure",
-                "azure_openai": "azure",
-                "google": "gemini",
-                "gemini": "gemini",
-                "bedrock": "bedrock",
-                "aws": "bedrock",
-            }
-
-            canonical_provider = provider_mapping.get(prefix.lower())
-
-            if canonical_provider and cls._validate_model_in_constants(
-                model_part, canonical_provider
-            ):
-                provider = canonical_provider
-                use_native = True
-                model_string = model_part
-            else:
-                provider = prefix
-                use_native = False
-                model_string = model_part
-        else:
-            provider = cls._infer_provider_from_model(model)
-            use_native = True
-            model_string = model
-
-        native_class = cls._get_native_provider(provider) if use_native else None
-        if native_class and not is_litellm and provider in SUPPORTED_NATIVE_PROVIDERS:
-            try:
-                # Remove 'provider' from kwargs if it exists to avoid duplicate keyword argument
-                kwargs_copy = {k: v for k, v in kwargs.items() if k != 'provider'}
-                return cast(
-                    Self, native_class(model=model_string, provider=provider, **kwargs_copy)
-                )
-            except NotImplementedError:
-                raise
-            except Exception as e:
-                raise ImportError(f"Error importing native provider: {e}") from e
-
-        # FALLBACK to LiteLLM
-        if not LITELLM_AVAILABLE:
-            logger.error("LiteLLM is not available, falling back to LiteLLM")
-            raise ImportError("Fallback to LiteLLM is not available") from None
-
-        instance = object.__new__(cls)
-        super(LLM, instance).__init__(model=model, is_litellm=True, **kwargs)
-        instance.is_litellm = True
-        return instance
-
-    @classmethod
-    def _validate_model_in_constants(cls, model: str, provider: str) -> bool:
-        """Validate if a model name exists in the provider's constants.
+    def model_post_init(self, __context: Any) -> None:
+        """Initialize LiteLLM-specific settings after model initialization.
 
         Args:
-            model: The model name to validate
-            provider: The provider to check against (canonical name)
-
-        Returns:
-            True if the model exists in the provider's constants, False otherwise
+            __context: Pydantic context
         """
-        if provider == "openai":
-            return model in OPENAI_MODELS
+        super().model_post_init(__context)
 
-        if provider == "anthropic" or provider == "claude":
-            return model in ANTHROPIC_MODELS
+        # Configure LiteLLM
+        if LITELLM_AVAILABLE:
+            litellm.drop_params = True
 
-        if provider == "gemini":
-            return model in GEMINI_MODELS
+        # Determine if this is an Anthropic model
+        self.is_anthropic = self._is_anthropic_model(self.model)
 
-        if provider == "bedrock":
-            return model in BEDROCK_MODELS
-
-        if provider == "azure":
-            # azure does not provide a list of available models, determine a better way to handle this
-            return True
-
-        return False
-
-    @classmethod
-    def _infer_provider_from_model(cls, model: str) -> str:
-        """Infer the provider from the model name.
-
-        Args:
-            model: The model name without provider prefix
-
-        Returns:
-            The inferred provider name, defaults to "openai"
-        """
-
-        if model in OPENAI_MODELS:
-            return "openai"
-
-        if model in ANTHROPIC_MODELS:
-            return "anthropic"
-
-        if model in GEMINI_MODELS:
-            return "gemini"
-
-        if model in BEDROCK_MODELS:
-            return "bedrock"
-
-        if model in AZURE_MODELS:
-            return "azure"
-
-        return "openai"
-
-    @classmethod
-    def _get_native_provider(cls, provider: str) -> type | None:
-        """Get native provider class if available."""
-        if provider == "openai":
-            from crewai.llms.providers.openai.completion import OpenAICompletion
-
-            return OpenAICompletion
-
-        if provider == "anthropic" or provider == "claude":
-            from crewai.llms.providers.anthropic.completion import (
-                AnthropicCompletion,
-            )
-
-            return AnthropicCompletion
-
-        if provider == "azure" or provider == "azure_openai":
-            from crewai.llms.providers.azure.completion import AzureCompletion
-
-            return AzureCompletion
-
-        if provider == "google" or provider == "gemini":
-            from crewai.llms.providers.gemini.completion import GeminiCompletion
-
-            return GeminiCompletion
-
-        if provider == "bedrock":
-            from crewai.llms.providers.bedrock.completion import BedrockCompletion
-
-            return BedrockCompletion
-
-        return None
-
-    def __init__(
-        self,
-        model: str,
-        timeout: float | int | None = None,
-        temperature: float | None = None,
-        top_p: float | None = None,
-        n: int | None = None,
-        stop: str | list[str] | None = None,
-        max_completion_tokens: int | None = None,
-        max_tokens: int | float | None = None,
-        presence_penalty: float | None = None,
-        frequency_penalty: float | None = None,
-        logit_bias: dict[int, float] | None = None,
-        response_format: type[BaseModel] | None = None,
-        seed: int | None = None,
-        logprobs: int | None = None,
-        top_logprobs: int | None = None,
-        base_url: str | None = None,
-        api_base: str | None = None,
-        api_version: str | None = None,
-        api_key: str | None = None,
-        callbacks: list[Any] | None = None,
-        reasoning_effort: Literal["none", "low", "medium", "high"] | None = None,
-        stream: bool = False,
-        interceptor: BaseInterceptor[httpx.Request, httpx.Response] | None = None,
-        **kwargs: Any,
-    ) -> None:
-        """Initialize LLM instance.
-
-        Note: This __init__ method is only called for fallback instances.
-        Native provider instances handle their own initialization in their respective classes.
-        """
-        super().__init__(
-            model=model,
-            temperature=temperature,
-            api_key=api_key,
-            base_url=base_url,
-            timeout=timeout,
-            **kwargs,
-        )
-        self.model = model
-        self.timeout = timeout
-        self.temperature = temperature
-        self.top_p = top_p
-        self.n = n
-        self.max_completion_tokens = max_completion_tokens
-        self.max_tokens = max_tokens
-        self.presence_penalty = presence_penalty
-        self.frequency_penalty = frequency_penalty
-        self.logit_bias = logit_bias
-        self.response_format = response_format
-        self.seed = seed
-        self.logprobs = logprobs
-        self.top_logprobs = top_logprobs
-        self.base_url = base_url
-        self.api_base = api_base
-        self.api_version = api_version
-        self.api_key = api_key
-        self.callbacks = callbacks
-        self.context_window_size = 0
-        self.reasoning_effort = reasoning_effort
-        self.additional_params = kwargs
-        self.is_anthropic = self._is_anthropic_model(model)
-        self.stream = stream
-        self.interceptor = interceptor
-
-        litellm.drop_params = True
-
-        # Normalize self.stop to always be a list[str]
-        if stop is None:
-            self.stop: list[str] = []
-        elif isinstance(stop, str):
-            self.stop = [stop]
-        else:
-            self.stop = stop
-
-        self.set_callbacks(callbacks or [])
+        # Set up callbacks
+        self.set_callbacks(self.callbacks or [])
         self.set_env_callbacks()
 
     @staticmethod
@@ -1649,7 +1447,7 @@ class LLM(BaseLLM):
             **filtered_params,
         )
 
-    def __deepcopy__(self, memo: dict[int, Any] | None) -> LLM:
+    def __deepcopy__(self, memo: dict[int, Any] | None) -> LLM:  # type: ignore[override]
         """Create a deep copy of the LLM instance."""
         import copy
 

lib/crewai/src/crewai/llm/hooks/__init__.py

@@ -0,0 +1,6 @@
+"""Interceptor contracts for crewai"""
+
+from crewai.llm.hooks.base import BaseInterceptor
+
+
+__all__ = ["BaseInterceptor"]

lib/crewai/src/crewai/llm/hooks/base.py

@@ -0,0 +1,133 @@
+"""Base classes for LLM transport interceptors.
+
+This module provides abstract base classes for intercepting and modifying
+outbound and inbound messages at the transport level.
+"""
+
+from __future__ import annotations
+
+from abc import ABC, abstractmethod
+from typing import TYPE_CHECKING, Any, Generic, TypeVar
+
+from pydantic_core import core_schema
+
+
+if TYPE_CHECKING:
+    from pydantic import GetCoreSchemaHandler
+    from pydantic_core import CoreSchema
+
+
+T = TypeVar("T")
+U = TypeVar("U")
+
+
+class BaseInterceptor(ABC, Generic[T, U]):
+    """Abstract base class for intercepting transport-level messages.
+
+    Provides hooks to intercept and modify outbound and inbound messages
+    at the transport layer.
+
+    Type parameters:
+        T: Outbound message type (e.g., httpx.Request)
+        U: Inbound message type (e.g., httpx.Response)
+
+    Example:
+        >>> import httpx
+        >>> class CustomInterceptor(BaseInterceptor[httpx.Request, httpx.Response]):
+        ...     def on_outbound(self, message: httpx.Request) -> httpx.Request:
+        ...         message.headers["X-Custom-Header"] = "value"
+        ...         return message
+        ...
+        ...     def on_inbound(self, message: httpx.Response) -> httpx.Response:
+        ...         print(f"Status: {message.status_code}")
+        ...         return message
+    """
+
+    @abstractmethod
+    def on_outbound(self, message: T) -> T:
+        """Intercept outbound message before sending.
+
+        Args:
+            message: Outbound message object.
+
+        Returns:
+            Modified message object.
+        """
+        ...
+
+    @abstractmethod
+    def on_inbound(self, message: U) -> U:
+        """Intercept inbound message after receiving.
+
+        Args:
+            message: Inbound message object.
+
+        Returns:
+            Modified message object.
+        """
+        ...
+
+    async def aon_outbound(self, message: T) -> T:
+        """Async version of on_outbound.
+
+        Args:
+            message: Outbound message object.
+
+        Returns:
+            Modified message object.
+        """
+        raise NotImplementedError
+
+    async def aon_inbound(self, message: U) -> U:
+        """Async version of on_inbound.
+
+        Args:
+            message: Inbound message object.
+
+        Returns:
+            Modified message object.
+        """
+        raise NotImplementedError
+
+    @classmethod
+    def __get_pydantic_core_schema__(
+        cls, _source_type: Any, _handler: GetCoreSchemaHandler
+    ) -> CoreSchema:
+        """Generate Pydantic core schema for BaseInterceptor.
+
+        This allows the generic BaseInterceptor to be used in Pydantic models
+        without requiring arbitrary_types_allowed=True. The schema validates
+        that the value is an instance of BaseInterceptor.
+
+        Args:
+            _source_type: The source type being validated (unused).
+            _handler: Handler for generating schemas (unused).
+
+        Returns:
+            A Pydantic core schema that validates BaseInterceptor instances.
+        """
+        return core_schema.no_info_plain_validator_function(
+            _validate_interceptor,
+            serialization=core_schema.plain_serializer_function_ser_schema(
+                lambda x: x, return_schema=core_schema.any_schema()
+            ),
+        )
+
+
+def _validate_interceptor(value: Any) -> BaseInterceptor[T, U]:
+    """Validate that the value is a BaseInterceptor instance.
+
+    Args:
+        value: The value to validate.
+
+    Returns:
+        The validated BaseInterceptor instance.
+
+    Raises:
+        ValueError: If the value is not a BaseInterceptor instance.
+    """
+    if not isinstance(value, BaseInterceptor):
+        raise ValueError(
+            f"Expected BaseInterceptor instance, got {type(value).__name__}"
+        )
+    return value

lib/crewai/src/crewai/llm/hooks/transport.py

@@ -0,0 +1,123 @@
+"""HTTP transport implementations for LLM request/response interception.
+
+This module provides internal transport classes that integrate with BaseInterceptor
+to enable request/response modification at the transport level.
+"""
+
+from __future__ import annotations
+
+from collections.abc import Iterable
+from typing import TYPE_CHECKING, TypedDict
+
+from httpx import (
+    AsyncHTTPTransport as _AsyncHTTPTransport,
+    HTTPTransport as _HTTPTransport,
+)
+from typing_extensions import NotRequired, Unpack
+
+
+if TYPE_CHECKING:
+    from ssl import SSLContext
+
+    from httpx import Limits, Request, Response
+    from httpx._types import CertTypes, ProxyTypes
+
+    from crewai.llm.hooks.base import BaseInterceptor
+
+
+class HTTPTransportKwargs(TypedDict, total=False):
+    """Typed dictionary for httpx.HTTPTransport initialization parameters.
+
+    These parameters configure the underlying HTTP transport behavior including
+    SSL verification, proxies, connection limits, and low-level socket options.
+    """
+
+    verify: bool | str | SSLContext
+    cert: NotRequired[CertTypes]
+    trust_env: bool
+    http1: bool
+    http2: bool
+    limits: Limits
+    proxy: NotRequired[ProxyTypes]
+    uds: NotRequired[str]
+    local_address: NotRequired[str]
+    retries: int
+    socket_options: NotRequired[
+        Iterable[
+            tuple[int, int, int]
+            | tuple[int, int, bytes | bytearray]
+            | tuple[int, int, None, int]
+        ]
+    ]
+
+
+class HTTPTransport(_HTTPTransport):
+    """HTTP transport that uses an interceptor for request/response modification.
+
+    This transport is used internally when a user provides a BaseInterceptor.
+    Users should not instantiate this class directly - instead, pass an interceptor
+    to the LLM client and this transport will be created automatically.
+    """
+
+    def __init__(
+        self,
+        interceptor: BaseInterceptor[Request, Response],
+        **kwargs: Unpack[HTTPTransportKwargs],
+    ) -> None:
+        """Initialize transport with interceptor.
+
+        Args:
+            interceptor: HTTP interceptor for modifying raw request/response objects.
+            **kwargs: HTTPTransport configuration parameters (verify, cert, proxy, etc.).
+        """
+        super().__init__(**kwargs)
+        self.interceptor = interceptor
+
+    def handle_request(self, request: Request) -> Response:
+        """Handle request with interception.
+
+        Args:
+            request: The HTTP request to handle.
+
+        Returns:
+            The HTTP response.
+        """
+        request = self.interceptor.on_outbound(request)
+        response = super().handle_request(request)
+        return self.interceptor.on_inbound(response)
+
+
+class AsyncHTTPTransport(_AsyncHTTPTransport):
+    """Async HTTP transport that uses an interceptor for request/response modification.
+
+    This transport is used internally when a user provides a BaseInterceptor.
+    Users should not instantiate this class directly - instead, pass an interceptor
+    to the LLM client and this transport will be created automatically.
+    """
+
+    def __init__(
+        self,
+        interceptor: BaseInterceptor[Request, Response],
+        **kwargs: Unpack[HTTPTransportKwargs],
+    ) -> None:
+        """Initialize async transport with interceptor.
+
+        Args:
+            interceptor: HTTP interceptor for modifying raw request/response objects.
+            **kwargs: HTTPTransport configuration parameters (verify, cert, proxy, etc.).
+        """
+        super().__init__(**kwargs)
+        self.interceptor = interceptor
+
+    async def handle_async_request(self, request: Request) -> Response:
+        """Handle async request with interception.
+
+        Args:
+            request: The HTTP request to handle.
+
+        Returns:
+            The HTTP response.
+        """
+        request = await self.interceptor.aon_outbound(request)
+        response = await super().handle_async_request(request)
+        return await self.interceptor.aon_inbound(response)

lib/crewai/src/crewai/llm/internal/constants.py

@@ -0,0 +1,14 @@
+from crewai.llm.constants import SupportedNativeProviders
+
+
+PROVIDER_MAPPING: dict[str, SupportedNativeProviders] = {
+    "openai": "openai",
+    "anthropic": "anthropic",
+    "claude": "anthropic",
+    "azure": "azure",
+    "azure_openai": "azure",
+    "google": "gemini",
+    "gemini": "gemini",
+    "bedrock": "bedrock",
+    "aws": "bedrock",
+}

lib/crewai/src/crewai/llm/internal/meta.py

@@ -0,0 +1,251 @@
+"""Metaclass for LLM provider routing.
+
+This metaclass enables automatic routing to native provider implementations
+based on the model parameter at instantiation time.
+"""
+
+from __future__ import annotations
+
+import logging
+from typing import Any, cast
+
+from pydantic import ConfigDict
+from pydantic._internal._model_construction import ModelMetaclass
+
+from crewai.llm.constants import (
+    ANTHROPIC_MODELS,
+    AZURE_MODELS,
+    BEDROCK_MODELS,
+    GEMINI_MODELS,
+    OPENAI_MODELS,
+    SUPPORTED_NATIVE_PROVIDERS,
+    SupportedModels,
+    SupportedNativeProviders,
+)
+from crewai.llm.internal.constants import PROVIDER_MAPPING
+
+
+class LLMMeta(ModelMetaclass):
+    """Metaclass for LLM that handles provider routing.
+
+    This metaclass intercepts LLM instantiation and routes to the appropriate
+    native provider implementation based on the model parameter.
+    """
+
+    def __new__(
+        mcs,
+        name: str,
+        bases: tuple[type, ...],
+        namespace: dict[str, Any],
+        **kwargs: Any,
+    ) -> type:
+        """Create new LLM class with proper model_config for custom LLMs.
+
+        Args:
+            name: Class name
+            bases: Base classes
+            namespace: Class namespace
+            **kwargs: Additional arguments
+
+        Returns:
+            New class
+        """
+        if name != "BaseLLM" and any(
+            base.__name__ in ("BaseLLM", "LLM") for base in bases
+        ):
+            if "model_config" not in namespace:
+                namespace["model_config"] = ConfigDict(
+                    extra="allow", populate_by_name=True
+                )
+            elif isinstance(namespace["model_config"], dict):
+                config_dict = cast(
+                    ConfigDict, cast(object, dict(namespace["model_config"]))
+                )
+                config_dict.setdefault("extra", "allow")
+                config_dict.setdefault("populate_by_name", True)
+                namespace["model_config"] = ConfigDict(**config_dict)
+
+        return super().__new__(mcs, name, bases, namespace)
+
+    def __call__(cls, *args: Any, **kwargs: Any) -> Any:  # noqa: N805
+        """Route to appropriate provider implementation at instantiation time.
+
+        Args:
+            *args: Positional arguments (model should be first for LLM class)
+            **kwargs: Keyword arguments including model, is_litellm, etc.
+
+        Returns:
+            Instance of the appropriate provider class or LLM class
+
+        Raises:
+            ValueError: If model is not a valid string
+        """
+        if cls.__name__ != "LLM":
+            return super().__call__(*args, **kwargs)
+
+        model = cast(
+            str | SupportedModels | None,
+            (kwargs.get("model") or (args[0] if args else None)),
+        )
+        is_litellm = kwargs.get("is_litellm", False)
+
+        if not model or not isinstance(model, str):
+            raise ValueError("Model must be a non-empty string")
+
+        if args and not kwargs.get("model"):
+            kwargs["model"] = cast(SupportedModels, args[0])
+            _ = args[1:]
+        explicit_provider = cast(SupportedNativeProviders, kwargs.get("provider"))
+
+        if explicit_provider:
+            provider = explicit_provider
+            use_native = True
+            model_string = model
+        elif "/" in model:
+            prefix, _, model_part = cast(
+                tuple[SupportedNativeProviders, Any, SupportedModels],
+                model.partition("/"),
+            )
+
+            canonical_provider = PROVIDER_MAPPING.get(prefix.lower())
+
+            if canonical_provider and cls._validate_model_in_constants(
+                model_part, canonical_provider
+            ):
+                provider = canonical_provider
+                use_native = True
+                model_string = model_part
+            else:
+                provider = prefix
+                use_native = False
+                model_string = model_part
+        else:
+            provider = cls._infer_provider_from_model(model)
+            use_native = True
+            model_string = model
+
+        native_class = cls._get_native_provider(provider) if use_native else None
+        if native_class and not is_litellm and provider in SUPPORTED_NATIVE_PROVIDERS:
+            try:
+                kwargs_copy = {
+                    k: v for k, v in kwargs.items() if k not in ("provider", "model")
+                }
+                return native_class(
+                    model=model_string, provider=provider, **kwargs_copy
+                )
+            except NotImplementedError:
+                raise
+            except Exception as e:
+                raise ImportError(f"Error importing native provider: {e}") from e
+
+        try:
+            import litellm  # noqa: F401
+        except ImportError:
+            logging.error("LiteLLM is not available, falling back to LiteLLM")
+            raise ImportError("Fallback to LiteLLM is not available") from None
+
+        kwargs_copy = {
+            k: v for k, v in kwargs.items() if k not in ("model", "is_litellm")
+        }
+        return super().__call__(model=model, is_litellm=True, **kwargs_copy)
+
+    @staticmethod
+    def _validate_model_in_constants(
+        model: SupportedModels, provider: SupportedNativeProviders | None
+    ) -> bool:
+        """Validate if a model name exists in the provider's constants.
+
+        Args:
+            model: The model name to validate
+            provider: The provider to check against (canonical name)
+
+        Returns:
+            True if the model exists in the provider's constants, False otherwise
+        """
+
+        if provider == "openai":
+            return model in OPENAI_MODELS
+
+        if provider == "anthropic" or provider == "claude":
+            return model in ANTHROPIC_MODELS
+
+        if provider == "gemini":
+            return model in GEMINI_MODELS
+
+        if provider == "bedrock":
+            return model in BEDROCK_MODELS
+
+        if provider == "azure":
+            # azure does not provide a list of available models
+            return True
+
+        return False
+
+    @staticmethod
+    def _infer_provider_from_model(
+        model: SupportedModels | str,
+    ) -> SupportedNativeProviders:
+        """Infer the provider from the model name.
+
+        Args:
+            model: The model name without provider prefix
+
+        Returns:
+            The inferred provider name, defaults to "openai"
+        """
+
+        if model in OPENAI_MODELS:
+            return "openai"
+
+        if model in ANTHROPIC_MODELS:
+            return "anthropic"
+
+        if model in GEMINI_MODELS:
+            return "gemini"
+
+        if model in BEDROCK_MODELS:
+            return "bedrock"
+
+        if model in AZURE_MODELS:
+            return "azure"
+
+        return "openai"
+
+    @staticmethod
+    def _get_native_provider(provider: SupportedNativeProviders | None) -> type | None:
+        """Get native provider class if available.
+
+        Args:
+            provider: The provider name
+
+        Returns:
+            The provider class or None if not available
+        """
+        if provider == "openai":
+            from crewai.llm.providers.openai.completion import OpenAICompletion
+
+            return OpenAICompletion
+
+        if provider == "anthropic" or provider == "claude":
+            from crewai.llm.providers.anthropic.completion import (
+                AnthropicCompletion,
+            )
+
+            return AnthropicCompletion
+
+        if provider == "azure" or provider == "azure_openai":
+            from crewai.llm.providers.azure.completion import AzureCompletion
+
+            return AzureCompletion
+
+        if provider == "google" or provider == "gemini":
+            from crewai.llm.providers.gemini.completion import GeminiCompletion
+
+            return GeminiCompletion
+
+        if provider == "bedrock":
+            from crewai.llm.providers.bedrock.completion import BedrockCompletion
+
+            return BedrockCompletion
+
+        return None

lib/crewai/src/crewai/llm/providers/anthropic/completion.py

@@ -2,14 +2,18 @@ from __future__ import annotations
 
 import json
 import logging
-import os
 from typing import TYPE_CHECKING, Any, cast
 
-from pydantic import BaseModel
+from dotenv import load_dotenv
+import httpx
+from pydantic import BaseModel, Field, PrivateAttr, model_validator
+from typing_extensions import Self
 
 from crewai.events.types.llm_events import LLMCallType
-from crewai.llms.base_llm import BaseLLM
-from crewai.llms.hooks.transport import HTTPTransport
+from crewai.llm.base_llm import BaseLLM
+from crewai.llm.core import CONTEXT_WINDOW_USAGE_RATIO
+from crewai.llm.hooks.transport import HTTPTransport
+from crewai.llm.providers.utils.common import safe_tool_conversion
 from crewai.utilities.agent_utils import is_context_length_exceeded
 from crewai.utilities.exceptions.context_window_exceeding_exception import (
     LLMContextLengthExceededError,
@@ -18,114 +22,85 @@ from crewai.utilities.types import LLMMessage
 
 
 if TYPE_CHECKING:
-    from crewai.llms.hooks.base import BaseInterceptor
+    from anthropic.types import Message
+
+    from crewai.agent.core import Agent
+    from crewai.task import Task
+
 
 try:
     from anthropic import Anthropic
-    from anthropic.types import Message
     from anthropic.types.tool_use_block import ToolUseBlock
-    import httpx
 except ImportError:
     raise ImportError(
         'Anthropic native provider not available, to install: uv add "crewai[anthropic]"'
     ) from None
 
 
+load_dotenv()
+
+
 class AnthropicCompletion(BaseLLM):
     """Anthropic native completion implementation.
 
     This class provides direct integration with the Anthropic Python SDK,
     offering native tool use, streaming support, and proper message formatting.
+
+    Attributes:
+        model: Anthropic model name (e.g., 'claude-3-5-sonnet-20241022')
+        base_url: Custom base URL for Anthropic API
+        timeout: Request timeout in seconds
+        max_retries: Maximum number of retries
+        max_tokens: Maximum tokens in response (required for Anthropic)
+        top_p: Nucleus sampling parameter
+        stream: Enable streaming responses
+        client_params: Additional parameters for the Anthropic client
+        interceptor: HTTP interceptor for modifying requests/responses at transport level
     """
 
-    def __init__(
-        self,
-        model: str = "claude-3-5-sonnet-20241022",
-        api_key: str | None = None,
-        base_url: str | None = None,
-        timeout: float | None = None,
-        max_retries: int = 2,
-        temperature: float | None = None,
-        max_tokens: int = 4096,  # Required for Anthropic
-        top_p: float | None = None,
-        stop_sequences: list[str] | None = None,
-        stream: bool = False,
-        client_params: dict[str, Any] | None = None,
-        interceptor: BaseInterceptor[httpx.Request, httpx.Response] | None = None,
-        **kwargs: Any,
-    ):
-        """Initialize Anthropic chat completion client.
+    base_url: str | None = Field(
+        default=None, description="Custom base URL for Anthropic API"
+    )
+    timeout: float | None = Field(
+        default=None, description="Request timeout in seconds"
+    )
+    max_retries: int = Field(default=2, description="Maximum number of retries")
+    max_tokens: int = Field(
+        default=4096, description="Maximum tokens in response (required for Anthropic)"
+    )
+    top_p: float | None = Field(default=None, description="Nucleus sampling parameter")
+    stream: bool = Field(default=False, description="Enable streaming responses")
+    client_params: dict[str, Any] | None = Field(
+        default_factory=dict, description="Additional Anthropic client parameters"
+    )
+    _client: Anthropic = PrivateAttr(default=None)  # type: ignore[assignment]
 
-        Args:
-            model: Anthropic model name (e.g., 'claude-3-5-sonnet-20241022')
-            api_key: Anthropic API key (defaults to ANTHROPIC_API_KEY env var)
-            base_url: Custom base URL for Anthropic API
-            timeout: Request timeout in seconds
-            max_retries: Maximum number of retries
-            temperature: Sampling temperature (0-1)
-            max_tokens: Maximum tokens in response (required for Anthropic)
-            top_p: Nucleus sampling parameter
-            stop_sequences: Stop sequences (Anthropic uses stop_sequences, not stop)
-            stream: Enable streaming responses
-            client_params: Additional parameters for the Anthropic client
-            interceptor: HTTP interceptor for modifying requests/responses at transport level.
-            **kwargs: Additional parameters
-        """
-        super().__init__(
-            model=model, temperature=temperature, stop=stop_sequences or [], **kwargs
-        )
+    _is_claude_3: bool = PrivateAttr(default=False)
+    _supports_tools: bool = PrivateAttr(default=False)
 
-        # Client params
-        self.interceptor = interceptor
-        self.client_params = client_params
-        self.base_url = base_url
-        self.timeout = timeout
-        self.max_retries = max_retries
+    @model_validator(mode="after")
+    def setup_client(self) -> Self:
+        """Initialize the Anthropic client and model-specific settings."""
+        self._client = Anthropic(**self._get_client_params())
 
-        self.client = Anthropic(**self._get_client_params())
+        self._is_claude_3 = "claude-3" in self.model.lower()
+        self._supports_tools = self._is_claude_3
 
-        # Store completion parameters
-        self.max_tokens = max_tokens
-        self.top_p = top_p
-        self.stream = stream
-        self.stop_sequences = stop_sequences or []
-
-        # Model-specific settings
-        self.is_claude_3 = "claude-3" in model.lower()
-        self.supports_tools = self.is_claude_3  # Claude 3+ supports tool use
+        return self
 
     @property
-    def stop(self) -> list[str]:
-        """Get stop sequences sent to the API."""
-        return self.stop_sequences
+    def is_claude_3(self) -> bool:
+        """Check if model is Claude 3."""
+        return self._is_claude_3
 
-    @stop.setter
-    def stop(self, value: list[str] | str | None) -> None:
-        """Set stop sequences.
-
-        Synchronizes stop_sequences to ensure values set by CrewAgentExecutor
-        are properly sent to the Anthropic API.
-
-        Args:
-            value: Stop sequences as a list, single string, or None
-        """
-        if value is None:
-            self.stop_sequences = []
-        elif isinstance(value, str):
-            self.stop_sequences = [value]
-        elif isinstance(value, list):
-            self.stop_sequences = value
-        else:
-            self.stop_sequences = []
+    @property
+    def supports_tools(self) -> bool:
+        """Check if model supports tools."""
+        return self._supports_tools
 
     def _get_client_params(self) -> dict[str, Any]:
         """Get client parameters."""
 
-        if self.api_key is None:
-            self.api_key = os.getenv("ANTHROPIC_API_KEY")
-            if self.api_key is None:
-                raise ValueError("ANTHROPIC_API_KEY is required")
-
         client_params = {
             "api_key": self.api_key,
             "base_url": self.base_url,
@@ -149,8 +124,8 @@ class AnthropicCompletion(BaseLLM):
         tools: list[dict[str, Any]] | None = None,
         callbacks: list[Any] | None = None,
         available_functions: dict[str, Any] | None = None,
-        from_task: Any | None = None,
-        from_agent: Any | None = None,
+        from_task: Task | None = None,
+        from_agent: Agent | None = None,
         response_model: type[BaseModel] | None = None,
     ) -> str | Any:
         """Call Anthropic messages API.
@@ -245,8 +220,8 @@ class AnthropicCompletion(BaseLLM):
             params["temperature"] = self.temperature
         if self.top_p is not None:
             params["top_p"] = self.top_p
-        if self.stop_sequences:
-            params["stop_sequences"] = self.stop_sequences
+        if self.stop:
+            params["stop_sequences"] = self.stop
 
         # Handle tools for Claude 3+
         if tools and self.supports_tools:
@@ -266,8 +241,6 @@ class AnthropicCompletion(BaseLLM):
                 continue
 
             try:
-                from crewai.llms.providers.utils.common import safe_tool_conversion
-
                 name, description, parameters = safe_tool_conversion(tool, "Anthropic")
             except (ImportError, KeyError, ValueError) as e:
                 logging.error(f"Error converting tool to Anthropic format: {e}")
@@ -341,8 +314,8 @@ class AnthropicCompletion(BaseLLM):
         self,
         params: dict[str, Any],
         available_functions: dict[str, Any] | None = None,
-        from_task: Any | None = None,
-        from_agent: Any | None = None,
+        from_task: Task | None = None,
+        from_agent: Agent | None = None,
         response_model: type[BaseModel] | None = None,
     ) -> str | Any:
         """Handle non-streaming message completion."""
@@ -357,7 +330,7 @@ class AnthropicCompletion(BaseLLM):
             params["tool_choice"] = {"type": "tool", "name": "structured_output"}
 
         try:
-            response: Message = self.client.messages.create(**params)
+            response: Message = self._client.messages.create(**params)
 
         except Exception as e:
             if is_context_length_exceeded(e):
@@ -429,8 +402,8 @@ class AnthropicCompletion(BaseLLM):
         self,
         params: dict[str, Any],
         available_functions: dict[str, Any] | None = None,
-        from_task: Any | None = None,
-        from_agent: Any | None = None,
+        from_task: Task | None = None,
+        from_agent: Agent | None = None,
         response_model: type[BaseModel] | None = None,
     ) -> str:
         """Handle streaming message completion."""
@@ -451,7 +424,7 @@ class AnthropicCompletion(BaseLLM):
         stream_params = {k: v for k, v in params.items() if k != "stream"}
 
         # Make streaming API call
-        with self.client.messages.stream(**stream_params) as stream:
+        with self._client.messages.stream(**stream_params) as stream:
             for event in stream:
                 if hasattr(event, "delta") and hasattr(event.delta, "text"):
                     text_delta = event.delta.text
@@ -525,8 +498,8 @@ class AnthropicCompletion(BaseLLM):
         tool_uses: list[ToolUseBlock],
         params: dict[str, Any],
         available_functions: dict[str, Any],
-        from_task: Any | None = None,
-        from_agent: Any | None = None,
+        from_task: Task | None = None,
+        from_agent: Agent | None = None,
     ) -> str:
         """Handle the complete tool use conversation flow.
 
@@ -579,7 +552,7 @@ class AnthropicCompletion(BaseLLM):
 
         try:
             # Send tool results back to Claude for final response
-            final_response: Message = self.client.messages.create(**follow_up_params)
+            final_response: Message = self._client.messages.create(**follow_up_params)
 
             # Track token usage for follow-up call
             follow_up_usage = self._extract_anthropic_token_usage(final_response)
@@ -636,7 +609,6 @@ class AnthropicCompletion(BaseLLM):
 
     def get_context_window_size(self) -> int:
         """Get the context window size for the model."""
-        from crewai.llm import CONTEXT_WINDOW_USAGE_RATIO
 
         # Context window sizes for Anthropic models
         context_windows = {

lib/crewai/src/crewai/llm/providers/azure/completion.py

@@ -5,8 +5,12 @@ import logging
 import os
 from typing import TYPE_CHECKING, Any
 
-from pydantic import BaseModel
+from dotenv import load_dotenv
+from pydantic import BaseModel, Field, PrivateAttr, model_validator
+from typing_extensions import Self
 
+from crewai.llm.core import CONTEXT_WINDOW_USAGE_RATIO, LLM_CONTEXT_WINDOW_SIZES
+from crewai.llm.providers.utils.common import safe_tool_conversion
 from crewai.utilities.agent_utils import is_context_length_exceeded
 from crewai.utilities.exceptions.context_window_exceeding_exception import (
     LLMContextLengthExceededError,
@@ -15,7 +19,8 @@ from crewai.utilities.types import LLMMessage
 
 
 if TYPE_CHECKING:
-    from crewai.llms.hooks.base import BaseInterceptor
+    from crewai.agent.core import Agent
+    from crewai.task import Task
     from crewai.tools.base_tool import BaseTool
 
 
@@ -36,7 +41,7 @@ try:
     )
 
     from crewai.events.types.llm_events import LLMCallType
-    from crewai.llms.base_llm import BaseLLM
+    from crewai.llm.base_llm import BaseLLM
 
 except ImportError:
     raise ImportError(
@@ -44,111 +49,109 @@ except ImportError:
     ) from None
 
 
+load_dotenv()
+
+
 class AzureCompletion(BaseLLM):
     """Azure AI Inference native completion implementation.
 
     This class provides direct integration with the Azure AI Inference Python SDK,
     offering native function calling, streaming support, and proper Azure authentication.
+
+    Attributes:
+        model: Azure deployment name or model name
+        endpoint: Azure endpoint URL
+        api_version: Azure API version
+        timeout: Request timeout in seconds
+        max_retries: Maximum number of retries
+        top_p: Nucleus sampling parameter
+        frequency_penalty: Frequency penalty (-2 to 2)
+        presence_penalty: Presence penalty (-2 to 2)
+        max_tokens: Maximum tokens in response
+        stream: Enable streaming responses
+        interceptor: HTTP interceptor (not yet supported for Azure)
     """
 
-    def __init__(
-        self,
-        model: str,
-        api_key: str | None = None,
-        endpoint: str | None = None,
-        api_version: str | None = None,
-        timeout: float | None = None,
-        max_retries: int = 2,
-        temperature: float | None = None,
-        top_p: float | None = None,
-        frequency_penalty: float | None = None,
-        presence_penalty: float | None = None,
-        max_tokens: int | None = None,
-        stop: list[str] | None = None,
-        stream: bool = False,
-        interceptor: BaseInterceptor[Any, Any] | None = None,
-        **kwargs: Any,
-    ):
-        """Initialize Azure AI Inference chat completion client.
+    endpoint: str = Field(  # type: ignore[assignment]
+        default_factory=lambda: os.getenv("AZURE_ENDPOINT")
+        or os.getenv("AZURE_OPENAI_ENDPOINT")
+        or os.getenv("AZURE_API_BASE"),
+        description="Azure endpoint URL (defaults to AZURE_ENDPOINT env var)",
+    )
+    api_version: str = Field(
+        default_factory=lambda: os.getenv("AZURE_API_VERSION", "2024-06-01"),
+        description="Azure API version (defaults to AZURE_API_VERSION env var or 2024-06-01)",
+    )
+    timeout: float | None = Field(
+        default=None, description="Request timeout in seconds"
+    )
+    max_retries: int = Field(default=2, description="Maximum number of retries")
+    top_p: float | None = Field(default=None, description="Nucleus sampling parameter")
+    frequency_penalty: float | None = Field(
+        default=None, le=2.0, ge=-2.0, description="Frequency penalty (-2 to 2)"
+    )
+    presence_penalty: float | None = Field(
+        default=None, le=2.0, ge=-2.0, description="Presence penalty (-2 to 2)"
+    )
+    max_tokens: int | None = Field(
+        default=None, description="Maximum tokens in response"
+    )
+    stream: bool = Field(default=False, description="Enable streaming responses")
+    _client: ChatCompletionsClient = PrivateAttr(default=None)  # type: ignore[assignment]
 
-        Args:
-            model: Azure deployment name or model name
-            api_key: Azure API key (defaults to AZURE_API_KEY env var)
-            endpoint: Azure endpoint URL (defaults to AZURE_ENDPOINT env var)
-            api_version: Azure API version (defaults to AZURE_API_VERSION env var)
-            timeout: Request timeout in seconds
-            max_retries: Maximum number of retries
-            temperature: Sampling temperature (0-2)
-            top_p: Nucleus sampling parameter
-            frequency_penalty: Frequency penalty (-2 to 2)
-            presence_penalty: Presence penalty (-2 to 2)
-            max_tokens: Maximum tokens in response
-            stop: Stop sequences
-            stream: Enable streaming responses
-            interceptor: HTTP interceptor (not yet supported for Azure).
-            **kwargs: Additional parameters
-        """
-        if interceptor is not None:
+    _is_openai_model: bool = PrivateAttr(default=False)
+    _is_azure_openai_endpoint: bool = PrivateAttr(default=False)
+
+    @model_validator(mode="after")
+    def setup_client(self) -> Self:
+        """Initialize the Azure client and validate configuration."""
+        if self.interceptor is not None:
             raise NotImplementedError(
                 "HTTP interceptors are not yet supported for Azure AI Inference provider. "
                 "Interceptors are currently supported for OpenAI and Anthropic providers only."
             )
 
-        super().__init__(
-            model=model, temperature=temperature, stop=stop or [], **kwargs
-        )
-
-        self.api_key = api_key or os.getenv("AZURE_API_KEY")
-        self.endpoint = (
-            endpoint
-            or os.getenv("AZURE_ENDPOINT")
-            or os.getenv("AZURE_OPENAI_ENDPOINT")
-            or os.getenv("AZURE_API_BASE")
-        )
-        self.api_version = api_version or os.getenv("AZURE_API_VERSION") or "2024-06-01"
-        self.timeout = timeout
-        self.max_retries = max_retries
+        if not self.api_key:
+            self.api_key = os.getenv("AZURE_API_KEY")
 
         if not self.api_key:
             raise ValueError(
                 "Azure API key is required. Set AZURE_API_KEY environment variable or pass api_key parameter."
             )
-        if not self.endpoint:
-            raise ValueError(
-                "Azure endpoint is required. Set AZURE_ENDPOINT environment variable or pass endpoint parameter."
-            )
 
-        # Validate and potentially fix Azure OpenAI endpoint URL
-        self.endpoint = self._validate_and_fix_endpoint(self.endpoint, model)
+        self.endpoint = self._validate_and_fix_endpoint(self.endpoint, self.model)
 
-        # Build client kwargs
-        client_kwargs = {
+        client_kwargs: dict[str, Any] = {
             "endpoint": self.endpoint,
             "credential": AzureKeyCredential(self.api_key),
         }
 
-        # Add api_version if specified (primarily for Azure OpenAI endpoints)
         if self.api_version:
             client_kwargs["api_version"] = self.api_version
 
-        self.client = ChatCompletionsClient(**client_kwargs)  # type: ignore[arg-type]
+        self._client = ChatCompletionsClient(**client_kwargs)
 
-        self.top_p = top_p
-        self.frequency_penalty = frequency_penalty
-        self.presence_penalty = presence_penalty
-        self.max_tokens = max_tokens
-        self.stream = stream
-
-        self.is_openai_model = any(
-            prefix in model.lower() for prefix in ["gpt-", "o1-", "text-"]
+        self._is_openai_model = any(
+            prefix in self.model.lower() for prefix in ["gpt-", "o1-", "text-"]
         )
-
-        self.is_azure_openai_endpoint = (
+        self._is_azure_openai_endpoint = (
             "openai.azure.com" in self.endpoint
             and "/openai/deployments/" in self.endpoint
         )
 
-    def _validate_and_fix_endpoint(self, endpoint: str, model: str) -> str:
+        return self
+
+    @property
+    def is_openai_model(self) -> bool:
+        """Check if model is an OpenAI model."""
+        return self._is_openai_model
+
+    @property
+    def is_azure_openai_endpoint(self) -> bool:
+        """Check if endpoint is an Azure OpenAI endpoint."""
+        return self._is_azure_openai_endpoint
+
+    def _validate_and_fix_endpoint(self, endpoint: str | None, model: str) -> str:
         """Validate and fix Azure endpoint URL format.
 
         Azure OpenAI endpoints should be in the format:
@@ -160,7 +163,15 @@ class AzureCompletion(BaseLLM):
 
         Returns:
             Validated and potentially corrected endpoint URL
+
+        Raises:
+            ValueError: If endpoint is None or empty
         """
+        if not endpoint:
+            raise ValueError(
+                "Azure endpoint is required. Set AZURE_ENDPOINT environment variable or pass endpoint parameter."
+            )
+
         if "openai.azure.com" in endpoint and "/openai/deployments/" not in endpoint:
             endpoint = endpoint.rstrip("/")
 
@@ -177,8 +188,8 @@ class AzureCompletion(BaseLLM):
         tools: list[dict[str, BaseTool]] | None = None,
         callbacks: list[Any] | None = None,
         available_functions: dict[str, Any] | None = None,
-        from_task: Any | None = None,
-        from_agent: Any | None = None,
+        from_task: Task | None = None,
+        from_agent: Agent | None = None,
         response_model: type[BaseModel] | None = None,
     ) -> str | Any:
         """Call Azure AI Inference chat completions API.
@@ -317,8 +328,6 @@ class AzureCompletion(BaseLLM):
     ) -> list[dict[str, Any]]:
         """Convert CrewAI tool format to Azure OpenAI function calling format."""
 
-        from crewai.llms.providers.utils.common import safe_tool_conversion
-
         azure_tools = []
 
         for tool in tools:
@@ -371,14 +380,14 @@ class AzureCompletion(BaseLLM):
         self,
         params: dict[str, Any],
         available_functions: dict[str, Any] | None = None,
-        from_task: Any | None = None,
-        from_agent: Any | None = None,
+        from_task: Task | None = None,
+        from_agent: Agent | None = None,
         response_model: type[BaseModel] | None = None,
     ) -> str | Any:
         """Handle non-streaming chat completion."""
         # Make API call
         try:
-            response: ChatCompletions = self.client.complete(**params)
+            response: ChatCompletions = self._client.complete(**params)
 
             if not response.choices:
                 raise ValueError("No choices returned from Azure API")
@@ -467,8 +476,8 @@ class AzureCompletion(BaseLLM):
         self,
         params: dict[str, Any],
         available_functions: dict[str, Any] | None = None,
-        from_task: Any | None = None,
-        from_agent: Any | None = None,
+        from_task: Task | None = None,
+        from_agent: Agent | None = None,
         response_model: type[BaseModel] | None = None,
     ) -> str:
         """Handle streaming chat completion."""
@@ -476,7 +485,7 @@ class AzureCompletion(BaseLLM):
         tool_calls = {}
 
         # Make streaming API call
-        for update in self.client.complete(**params):
+        for update in self._client.complete(**params):
             if isinstance(update, StreamingChatCompletionsUpdate):
                 if update.choices:
                     choice = update.choices[0]
@@ -554,7 +563,6 @@ class AzureCompletion(BaseLLM):
 
     def get_context_window_size(self) -> int:
         """Get the context window size for the model."""
-        from crewai.llm import CONTEXT_WINDOW_USAGE_RATIO, LLM_CONTEXT_WINDOW_SIZES
 
         min_context = 1024
         max_context = 2097152

lib/crewai/src/crewai/llm/providers/bedrock/completion.py

@@ -5,11 +5,15 @@ import logging
 import os
 from typing import TYPE_CHECKING, Any, TypedDict, cast
 
-from pydantic import BaseModel
-from typing_extensions import Required
+from dotenv import load_dotenv
+from mypy_boto3_bedrock_runtime.client import BedrockRuntimeClient
+from pydantic import BaseModel, Field, PrivateAttr, model_validator
+from typing_extensions import Required, Self
 
 from crewai.events.types.llm_events import LLMCallType
-from crewai.llms.base_llm import BaseLLM
+from crewai.llm.base_llm import BaseLLM
+from crewai.llm.core import CONTEXT_WINDOW_USAGE_RATIO
+from crewai.llm.providers.utils.common import safe_tool_conversion
 from crewai.utilities.agent_utils import is_context_length_exceeded
 from crewai.utilities.exceptions.context_window_exceeding_exception import (
     LLMContextLengthExceededError,
@@ -30,7 +34,8 @@ if TYPE_CHECKING:
         ToolTypeDef,
     )
 
-    from crewai.llms.hooks.base import BaseInterceptor
+    from crewai.agent.core import Agent
+    from crewai.task import Task
 
 
 try:
@@ -72,6 +77,9 @@ else:
         topK: int
 
 
+load_dotenv()
+
+
 class ToolInputSchema(TypedDict):
     """Type definition for tool input schema in Converse API."""
 
@@ -141,74 +149,84 @@ class BedrockCompletion(BaseLLM):
     - Complete streaming event handling (messageStart, contentBlockStart, etc.)
     - Response metadata and trace information capture
     - Model-specific conversation format handling (e.g., Cohere requirements)
+
+    Attributes:
+        model: The Bedrock model ID to use
+        aws_access_key_id: AWS access key (defaults to environment variable)
+        aws_secret_access_key: AWS secret key (defaults to environment variable)
+        aws_session_token: AWS session token for temporary credentials
+        region_name: AWS region name
+        max_tokens: Maximum tokens to generate
+        top_p: Nucleus sampling parameter
+        top_k: Top-k sampling parameter (Claude models only)
+        stream: Whether to use streaming responses
+        guardrail_config: Guardrail configuration for content filtering
+        additional_model_request_fields: Model-specific request parameters
+        additional_model_response_field_paths: Custom response field paths
+        interceptor: HTTP interceptor (not yet supported for Bedrock)
     """
 
-    def __init__(
-        self,
-        model: str = "anthropic.claude-3-5-sonnet-20241022-v2:0",
-        aws_access_key_id: str | None = None,
-        aws_secret_access_key: str | None = None,
-        aws_session_token: str | None = None,
-        region_name: str = "us-east-1",
-        temperature: float | None = None,
-        max_tokens: int | None = None,
-        top_p: float | None = None,
-        top_k: int | None = None,
-        stop_sequences: Sequence[str] | None = None,
-        stream: bool = False,
-        guardrail_config: dict[str, Any] | None = None,
-        additional_model_request_fields: dict[str, Any] | None = None,
-        additional_model_response_field_paths: list[str] | None = None,
-        interceptor: BaseInterceptor[Any, Any] | None = None,
-        **kwargs: Any,
-    ) -> None:
-        """Initialize AWS Bedrock completion client.
+    aws_access_key_id: str = Field(  # type: ignore[assignment]
+        default_factory=lambda: os.getenv("AWS_ACCESS_KEY_ID"),
+        description="AWS access key (defaults to environment variable)",
+    )
+    aws_secret_access_key: str = Field(  # type: ignore[assignment]
+        default_factory=lambda: os.getenv("AWS_SECRET_ACCESS_KEY"),
+        description="AWS secret key (defaults to environment variable)",
+    )
+    aws_session_token: str = Field(  # type: ignore[assignment]
+        default_factory=lambda: os.getenv("AWS_SESSION_TOKEN"),
+        description="AWS session token for temporary credentials",
+    )
+    region_name: str = Field(
+        default_factory=lambda: os.getenv("AWS_REGION", "us-east-1"),
+        description="AWS region name",
+    )
+    max_tokens: int | None = Field(
+        default=None, description="Maximum tokens to generate"
+    )
+    top_p: float | None = Field(default=None, description="Nucleus sampling parameter")
+    top_k: int | None = Field(
+        default=None, description="Top-k sampling parameter (Claude models only)"
+    )
+    stream: bool = Field(
+        default=False, description="Whether to use streaming responses"
+    )
+    guardrail_config: dict[str, Any] = Field(
+        default_factory=dict,
+        description="Guardrail configuration for content filtering",
+    )
+    additional_model_request_fields: dict[str, Any] = Field(
+        default_factory=dict, description="Model-specific request parameters"
+    )
+    additional_model_response_field_paths: list[str] = Field(
+        default_factory=list, description="Custom response field paths"
+    )
+    _client: BedrockRuntimeClient = PrivateAttr(  # type: ignore[assignment]
+        default_factory=lambda: Session().client,
+    )
 
-        Args:
-            model: The Bedrock model ID to use
-            aws_access_key_id: AWS access key (defaults to environment variable)
-            aws_secret_access_key: AWS secret key (defaults to environment variable)
-            aws_session_token: AWS session token for temporary credentials
-            region_name: AWS region name
-            temperature: Sampling temperature for response generation
-            max_tokens: Maximum tokens to generate
-            top_p: Nucleus sampling parameter
-            top_k: Top-k sampling parameter (Claude models only)
-            stop_sequences: List of sequences that stop generation
-            stream: Whether to use streaming responses
-            guardrail_config: Guardrail configuration for content filtering
-            additional_model_request_fields: Model-specific request parameters
-            additional_model_response_field_paths: Custom response field paths
-            interceptor: HTTP interceptor (not yet supported for Bedrock).
-            **kwargs: Additional parameters
-        """
-        if interceptor is not None:
+    _is_claude_model: bool = PrivateAttr(default=False)
+    _supports_tools: bool = PrivateAttr(default=True)
+    _supports_streaming: bool = PrivateAttr(default=True)
+    _model_id: str = PrivateAttr()
+
+    @model_validator(mode="after")
+    def setup_client(self) -> Self:
+        """Initialize the Bedrock client and validate configuration."""
+        if self.interceptor is not None:
             raise NotImplementedError(
                 "HTTP interceptors are not yet supported for AWS Bedrock provider. "
                 "Interceptors are currently supported for OpenAI and Anthropic providers only."
             )
 
-        # Extract provider from kwargs to avoid duplicate argument
-        kwargs.pop("provider", None)
-
-        super().__init__(
-            model=model,
-            temperature=temperature,
-            stop=stop_sequences or [],
-            provider="bedrock",
-            **kwargs,
-        )
-
-        # Initialize Bedrock client with proper configuration
         session = Session(
-            aws_access_key_id=aws_access_key_id or os.getenv("AWS_ACCESS_KEY_ID"),
-            aws_secret_access_key=aws_secret_access_key
-            or os.getenv("AWS_SECRET_ACCESS_KEY"),
-            aws_session_token=aws_session_token or os.getenv("AWS_SESSION_TOKEN"),
-            region_name=region_name,
+            aws_access_key_id=self.aws_access_key_id,
+            aws_secret_access_key=self.aws_secret_access_key,
+            aws_session_token=self.aws_session_token,
+            region_name=self.region_name,
         )
 
-        # Configure client with timeouts and retries following AWS best practices
         config = Config(
             read_timeout=300,
             retries={
@@ -218,54 +236,34 @@ class BedrockCompletion(BaseLLM):
             tcp_keepalive=True,
         )
 
-        self.client = session.client("bedrock-runtime", config=config)
-        self.region_name = region_name
+        self._client = session.client("bedrock-runtime", config=config)
 
-        # Store completion parameters
-        self.max_tokens = max_tokens
-        self.top_p = top_p
-        self.top_k = top_k
-        self.stream = stream
-        self.stop_sequences = stop_sequences or []
+        self._is_claude_model = "claude" in self.model.lower()
+        self._supports_tools = True
+        self._supports_streaming = True
+        self._model_id = self.model
 
-        # Store advanced features (optional)
-        self.guardrail_config = guardrail_config
-        self.additional_model_request_fields = additional_model_request_fields
-        self.additional_model_response_field_paths = (
-            additional_model_response_field_paths
-        )
-
-        # Model-specific settings
-        self.is_claude_model = "claude" in model.lower()
-        self.supports_tools = True  # Converse API supports tools for most models
-        self.supports_streaming = True
-
-        # Handle inference profiles for newer models
-        self.model_id = model
+        return self
 
     @property
-    def stop(self) -> list[str]:
-        """Get stop sequences sent to the API."""
-        return list(self.stop_sequences)
+    def is_claude_model(self) -> bool:
+        """Check if model is a Claude model."""
+        return self._is_claude_model
 
-    @stop.setter
-    def stop(self, value: Sequence[str] | str | None) -> None:
-        """Set stop sequences.
+    @property
+    def supports_tools(self) -> bool:
+        """Check if model supports tools."""
+        return self._supports_tools
 
-        Synchronizes stop_sequences to ensure values set by CrewAgentExecutor
-        are properly sent to the Bedrock API.
+    @property
+    def supports_streaming(self) -> bool:
+        """Check if model supports streaming."""
+        return self._supports_streaming
 
-        Args:
-            value: Stop sequences as a Sequence, single string, or None
-        """
-        if value is None:
-            self.stop_sequences = []
-        elif isinstance(value, str):
-            self.stop_sequences = [value]
-        elif isinstance(value, Sequence):
-            self.stop_sequences = list(value)
-        else:
-            self.stop_sequences = []
+    @property
+    def model_id(self) -> str:
+        """Get the model ID."""
+        return self._model_id
 
     def call(
         self,
@@ -273,8 +271,8 @@ class BedrockCompletion(BaseLLM):
         tools: list[dict[Any, Any]] | None = None,
         callbacks: list[Any] | None = None,
         available_functions: dict[str, Any] | None = None,
-        from_task: Any | None = None,
-        from_agent: Any | None = None,
+        from_task: Task | None = None,
+        from_agent: Agent | None = None,
         response_model: type[BaseModel] | None = None,
     ) -> str | Any:
         """Call AWS Bedrock Converse API."""
@@ -359,8 +357,8 @@ class BedrockCompletion(BaseLLM):
         messages: list[dict[str, Any]],
         body: BedrockConverseRequestBody,
         available_functions: Mapping[str, Any] | None = None,
-        from_task: Any | None = None,
-        from_agent: Any | None = None,
+        from_task: Task | None = None,
+        from_agent: Agent | None = None,
     ) -> str:
         """Handle non-streaming converse API call following AWS best practices."""
         try:
@@ -378,7 +376,7 @@ class BedrockCompletion(BaseLLM):
                     raise ValueError(f"Invalid message format at index {i}")
 
             # Call Bedrock Converse API with proper error handling
-            response = self.client.converse(
+            response = self._client.converse(
                 modelId=self.model_id,
                 messages=cast(
                     "Sequence[MessageTypeDef | MessageOutputTypeDef]",
@@ -540,8 +538,8 @@ class BedrockCompletion(BaseLLM):
         messages: list[dict[str, Any]],
         body: BedrockConverseRequestBody,
         available_functions: dict[str, Any] | None = None,
-        from_task: Any | None = None,
-        from_agent: Any | None = None,
+        from_task: Task | None = None,
+        from_agent: Agent | None = None,
     ) -> str:
         """Handle streaming converse API call with comprehensive event handling."""
         full_response = ""
@@ -549,7 +547,7 @@ class BedrockCompletion(BaseLLM):
         tool_use_id = None
 
         try:
-            response = self.client.converse_stream(
+            response = self._client.converse_stream(
                 modelId=self.model_id,
                 messages=cast(
                     "Sequence[MessageTypeDef | MessageOutputTypeDef]",
@@ -778,7 +776,6 @@ class BedrockCompletion(BaseLLM):
         tools: list[dict[str, Any]],
     ) -> list[ConverseToolTypeDef]:
         """Convert CrewAI tools to Converse API format following AWS specification."""
-        from crewai.llms.providers.utils.common import safe_tool_conversion
 
         converse_tools: list[ConverseToolTypeDef] = []
 
@@ -818,8 +815,8 @@ class BedrockCompletion(BaseLLM):
             config["temperature"] = float(self.temperature)
         if self.top_p is not None:
             config["topP"] = float(self.top_p)
-        if self.stop_sequences:
-            config["stopSequences"] = self.stop_sequences
+        if self.stop:
+            config["stopSequences"] = self.stop
 
         if self.is_claude_model and self.top_k is not None:
             # top_k is supported by Claude models
@@ -871,7 +868,6 @@ class BedrockCompletion(BaseLLM):
 
     def get_context_window_size(self) -> int:
         """Get the context window size for the model."""
-        from crewai.llm import CONTEXT_WINDOW_USAGE_RATIO
 
         # Context window sizes for common Bedrock models
         context_windows = {

lib/crewai/src/crewai/llm/providers/gemini/completion.py

@@ -1,12 +1,17 @@
+from __future__ import annotations
+
 import logging
 import os
-from typing import Any, cast
+from typing import TYPE_CHECKING, Any, cast
 
-from pydantic import BaseModel
+from dotenv import load_dotenv
+from pydantic import BaseModel, Field, PrivateAttr, model_validator
+from typing_extensions import Self
 
 from crewai.events.types.llm_events import LLMCallType
-from crewai.llms.base_llm import BaseLLM
-from crewai.llms.hooks.base import BaseInterceptor
+from crewai.llm.base_llm import BaseLLM
+from crewai.llm.core import CONTEXT_WINDOW_USAGE_RATIO, LLM_CONTEXT_WINDOW_SIZES
+from crewai.llm.providers.utils.common import safe_tool_conversion
 from crewai.utilities.agent_utils import is_context_length_exceeded
 from crewai.utilities.exceptions.context_window_exceeding_exception import (
     LLMContextLengthExceededError,
@@ -14,6 +19,11 @@ from crewai.utilities.exceptions.context_window_exceeding_exception import (
 from crewai.utilities.types import LLMMessage
 
 
+if TYPE_CHECKING:
+    from crewai.agent.core import Agent
+    from crewai.task import Task
+
+
 try:
     from google import genai  # type: ignore[import-untyped]
     from google.genai import types  # type: ignore[import-untyped]
@@ -24,111 +34,93 @@ except ImportError:
     ) from None
 
 
+load_dotenv()
+
+
 class GeminiCompletion(BaseLLM):
     """Google Gemini native completion implementation.
 
     This class provides direct integration with the Google Gen AI Python SDK,
     offering native function calling, streaming support, and proper Gemini formatting.
+
+    Attributes:
+        model: Gemini model name (e.g., 'gemini-2.0-flash-001', 'gemini-1.5-pro')
+        project: Google Cloud project ID (for Vertex AI)
+        location: Google Cloud location (for Vertex AI, defaults to 'us-central1')
+        top_p: Nucleus sampling parameter
+        top_k: Top-k sampling parameter
+        max_output_tokens: Maximum tokens in response
+        stream: Enable streaming responses
+        safety_settings: Safety filter settings
+        client_params: Additional parameters for Google Gen AI Client constructor
+        interceptor: HTTP interceptor (not yet supported for Gemini)
     """
 
-    def __init__(
-        self,
-        model: str = "gemini-2.0-flash-001",
-        api_key: str | None = None,
-        project: str | None = None,
-        location: str | None = None,
-        temperature: float | None = None,
-        top_p: float | None = None,
-        top_k: int | None = None,
-        max_output_tokens: int | None = None,
-        stop_sequences: list[str] | None = None,
-        stream: bool = False,
-        safety_settings: dict[str, Any] | None = None,
-        client_params: dict[str, Any] | None = None,
-        interceptor: BaseInterceptor[Any, Any] | None = None,
-        **kwargs: Any,
-    ):
-        """Initialize Google Gemini chat completion client.
+    project: str | None = Field(
+        default_factory=lambda: os.getenv("GOOGLE_CLOUD_PROJECT"),
+        description="Google Cloud project ID (for Vertex AI)",
+    )
+    location: str = Field(
+        default_factory=lambda: os.getenv("GOOGLE_CLOUD_LOCATION", "us-central1"),
+        description="Google Cloud location (for Vertex AI, defaults to 'us-central1')",
+    )
+    top_p: float | None = Field(default=None, description="Nucleus sampling parameter")
+    top_k: int | None = Field(default=None, description="Top-k sampling parameter")
+    max_output_tokens: int | None = Field(
+        default=None, description="Maximum tokens in response"
+    )
+    stream: bool = Field(default=False, description="Enable streaming responses")
+    safety_settings: dict[str, Any] = Field(
+        default_factory=dict, description="Safety filter settings"
+    )
+    client_params: dict[str, Any] = Field(
+        default_factory=dict,
+        description="Additional parameters for Google Gen AI Client constructor",
+    )
+    _client: Any = PrivateAttr(default=None)
 
-        Args:
-            model: Gemini model name (e.g., 'gemini-2.0-flash-001', 'gemini-1.5-pro')
-            api_key: Google API key (defaults to GOOGLE_API_KEY or GEMINI_API_KEY env var)
-            project: Google Cloud project ID (for Vertex AI)
-            location: Google Cloud location (for Vertex AI, defaults to 'us-central1')
-            temperature: Sampling temperature (0-2)
-            top_p: Nucleus sampling parameter
-            top_k: Top-k sampling parameter
-            max_output_tokens: Maximum tokens in response
-            stop_sequences: Stop sequences
-            stream: Enable streaming responses
-            safety_settings: Safety filter settings
-            client_params: Additional parameters to pass to the Google Gen AI Client constructor.
-                          Supports parameters like http_options, credentials, debug_config, etc.
-            interceptor: HTTP interceptor (not yet supported for Gemini).
-            **kwargs: Additional parameters
-        """
-        if interceptor is not None:
+    _is_gemini_2: bool = PrivateAttr(default=False)
+    _is_gemini_1_5: bool = PrivateAttr(default=False)
+    _supports_tools: bool = PrivateAttr(default=False)
+
+    @model_validator(mode="after")
+    def setup_client(self) -> Self:
+        """Initialize the Gemini client and validate configuration."""
+        if self.interceptor is not None:
             raise NotImplementedError(
                 "HTTP interceptors are not yet supported for Google Gemini provider. "
                 "Interceptors are currently supported for OpenAI and Anthropic providers only."
             )
 
-        super().__init__(
-            model=model, temperature=temperature, stop=stop_sequences or [], **kwargs
-        )
-
-        # Store client params for later use
-        self.client_params = client_params or {}
-
-        # Get API configuration with environment variable fallbacks
-        self.api_key = (
-            api_key or os.getenv("GOOGLE_API_KEY") or os.getenv("GEMINI_API_KEY")
-        )
-        self.project = project or os.getenv("GOOGLE_CLOUD_PROJECT")
-        self.location = location or os.getenv("GOOGLE_CLOUD_LOCATION") or "us-central1"
+        if self.api_key is None:
+            self.api_key = os.getenv("GOOGLE_API_KEY") or os.getenv("GEMINI_API_KEY")
 
         use_vertexai = os.getenv("GOOGLE_GENAI_USE_VERTEXAI", "").lower() == "true"
 
-        self.client = self._initialize_client(use_vertexai)
+        self._client = self._initialize_client(use_vertexai)
 
-        # Store completion parameters
-        self.top_p = top_p
-        self.top_k = top_k
-        self.max_output_tokens = max_output_tokens
-        self.stream = stream
-        self.safety_settings = safety_settings or {}
-        self.stop_sequences = stop_sequences or []
+        self._is_gemini_2 = "gemini-2" in self.model.lower()
+        self._is_gemini_1_5 = "gemini-1.5" in self.model.lower()
+        self._supports_tools = self._is_gemini_1_5 or self._is_gemini_2
 
-        # Model-specific settings
-        self.is_gemini_2 = "gemini-2" in model.lower()
-        self.is_gemini_1_5 = "gemini-1.5" in model.lower()
-        self.supports_tools = self.is_gemini_1_5 or self.is_gemini_2
+        return self
 
     @property
-    def stop(self) -> list[str]:
-        """Get stop sequences sent to the API."""
-        return self.stop_sequences
+    def is_gemini_2(self) -> bool:
+        """Check if model is Gemini 2."""
+        return self._is_gemini_2
 
-    @stop.setter
-    def stop(self, value: list[str] | str | None) -> None:
-        """Set stop sequences.
+    @property
+    def is_gemini_1_5(self) -> bool:
+        """Check if model is Gemini 1.5."""
+        return self._is_gemini_1_5
 
-        Synchronizes stop_sequences to ensure values set by CrewAgentExecutor
-        are properly sent to the Gemini API.
+    @property
+    def supports_tools(self) -> bool:
+        """Check if model supports tools."""
+        return self._supports_tools
 
-        Args:
-            value: Stop sequences as a list, single string, or None
-        """
-        if value is None:
-            self.stop_sequences = []
-        elif isinstance(value, str):
-            self.stop_sequences = [value]
-        elif isinstance(value, list):
-            self.stop_sequences = value
-        else:
-            self.stop_sequences = []
-
-    def _initialize_client(self, use_vertexai: bool = False) -> genai.Client:  # type: ignore[no-any-unimported]
+    def _initialize_client(self, use_vertexai: bool = False) -> Any:
         """Initialize the Google Gen AI client with proper parameter handling.
 
         Args:
@@ -150,12 +142,9 @@ class GeminiCompletion(BaseLLM):
                     "location": self.location,
                 }
             )
-
             client_params.pop("api_key", None)
-
         elif self.api_key:
             client_params["api_key"] = self.api_key
-
             client_params.pop("vertexai", None)
             client_params.pop("project", None)
             client_params.pop("location", None)
@@ -180,11 +169,10 @@ class GeminiCompletion(BaseLLM):
         params = {}
 
         if (
-            hasattr(self, "client")
-            and hasattr(self.client, "vertexai")
-            and self.client.vertexai
+            hasattr(self, "_client")
+            and hasattr(self._client, "vertexai")
+            and self._client.vertexai
         ):
-            # Vertex AI configuration
             params.update(
                 {
                     "vertexai": True,
@@ -206,8 +194,8 @@ class GeminiCompletion(BaseLLM):
         tools: list[dict[str, Any]] | None = None,
         callbacks: list[Any] | None = None,
         available_functions: dict[str, Any] | None = None,
-        from_task: Any | None = None,
-        from_agent: Any | None = None,
+        from_task: Task | None = None,
+        from_agent: Agent | None = None,
         response_model: type[BaseModel] | None = None,
     ) -> str | Any:
         """Call Google Gemini generate content API.
@@ -296,15 +284,12 @@ class GeminiCompletion(BaseLLM):
         self.tools = tools
         config_params = {}
 
-        # Add system instruction if present
         if system_instruction:
-            # Convert system instruction to Content format
             system_content = types.Content(
                 role="user", parts=[types.Part.from_text(text=system_instruction)]
             )
             config_params["system_instruction"] = system_content
 
-        # Add generation config parameters
         if self.temperature is not None:
             config_params["temperature"] = self.temperature
         if self.top_p is not None:
@@ -313,14 +298,13 @@ class GeminiCompletion(BaseLLM):
             config_params["top_k"] = self.top_k
         if self.max_output_tokens is not None:
             config_params["max_output_tokens"] = self.max_output_tokens
-        if self.stop_sequences:
-            config_params["stop_sequences"] = self.stop_sequences
+        if self.stop:
+            config_params["stop_sequences"] = self.stop
 
         if response_model:
             config_params["response_mime_type"] = "application/json"
             config_params["response_schema"] = response_model.model_json_schema()
 
-        # Handle tools for supported models
         if tools and self.supports_tools:
             config_params["tools"] = self._convert_tools_for_interference(tools)
 
@@ -335,8 +319,6 @@ class GeminiCompletion(BaseLLM):
         """Convert CrewAI tool format to Gemini function declaration format."""
         gemini_tools = []
 
-        from crewai.llms.providers.utils.common import safe_tool_conversion
-
         for tool in tools:
             name, description, parameters = safe_tool_conversion(tool, "Gemini")
 
@@ -345,7 +327,6 @@ class GeminiCompletion(BaseLLM):
                 description=description,
             )
 
-            # Add parameters if present - ensure parameters is a dict
             if parameters and isinstance(parameters, dict):
                 function_declaration.parameters = parameters
 
@@ -381,16 +362,12 @@ class GeminiCompletion(BaseLLM):
             content = message.get("content", "")
 
             if role == "system":
-                # Extract system instruction - Gemini handles it separately
                 if system_instruction:
                     system_instruction += f"\n\n{content}"
                 else:
                     system_instruction = cast(str, content)
             else:
-                # Convert role for Gemini (assistant -> model)
                 gemini_role = "model" if role == "assistant" else "user"
-
-                # Create Content object
                 gemini_content = types.Content(
                     role=gemini_role, parts=[types.Part.from_text(text=content)]
                 )
@@ -404,8 +381,8 @@ class GeminiCompletion(BaseLLM):
         system_instruction: str | None,
         config: types.GenerateContentConfig,
         available_functions: dict[str, Any] | None = None,
-        from_task: Any | None = None,
-        from_agent: Any | None = None,
+        from_task: Task | None = None,
+        from_agent: Agent | None = None,
         response_model: type[BaseModel] | None = None,
     ) -> str | Any:
         """Handle non-streaming content generation."""
@@ -416,7 +393,7 @@ class GeminiCompletion(BaseLLM):
         }
 
         try:
-            response = self.client.models.generate_content(**api_params)
+            response = self._client.models.generate_content(**api_params)
 
             usage = self._extract_token_usage(response)
         except Exception as e:
@@ -470,8 +447,8 @@ class GeminiCompletion(BaseLLM):
         contents: list[types.Content],
         config: types.GenerateContentConfig,
         available_functions: dict[str, Any] | None = None,
-        from_task: Any | None = None,
-        from_agent: Any | None = None,
+        from_task: Task | None = None,
+        from_agent: Agent | None = None,
         response_model: type[BaseModel] | None = None,
     ) -> str:
         """Handle streaming content generation."""
@@ -484,7 +461,7 @@ class GeminiCompletion(BaseLLM):
             "config": config,
         }
 
-        for chunk in self.client.models.generate_content_stream(**api_params):
+        for chunk in self._client.models.generate_content_stream(**api_params):
             if hasattr(chunk, "text") and chunk.text:
                 full_response += chunk.text
                 self._emit_stream_chunk_event(
@@ -507,13 +484,11 @@ class GeminiCompletion(BaseLLM):
                                     else {},
                                 }
 
-        # Handle completed function calls
         if function_calls and available_functions:
             for call_data in function_calls.values():
                 function_name = call_data["name"]
                 function_args = call_data["args"]
 
-                # Execute tool
                 result = self._handle_tool_execution(
                     function_name=function_name,
                     function_args=function_args,
@@ -547,7 +522,6 @@ class GeminiCompletion(BaseLLM):
 
     def get_context_window_size(self) -> int:
         """Get the context window size for the model."""
-        from crewai.llm import CONTEXT_WINDOW_USAGE_RATIO, LLM_CONTEXT_WINDOW_SIZES
 
         min_context = 1024
         max_context = 2097152
@@ -574,13 +548,11 @@ class GeminiCompletion(BaseLLM):
             "gemma-3-27b": 128000,
         }
 
-        # Find the best match for the model name
         for model_prefix, size in context_windows.items():
             if self.model.startswith(model_prefix):
                 return int(size * CONTEXT_WINDOW_USAGE_RATIO)
 
-        # Default context window size for Gemini models
-        return int(1048576 * CONTEXT_WINDOW_USAGE_RATIO)  # 1M tokens
+        return int(1048576 * CONTEXT_WINDOW_USAGE_RATIO)
 
     def _extract_token_usage(self, response: dict[str, Any]) -> dict[str, Any]:
         """Extract token usage from Gemini response."""

lib/crewai/src/crewai/llm/providers/openai/completion.py

@@ -6,16 +6,20 @@ import logging
 import os
 from typing import TYPE_CHECKING, Any
 
+from dotenv import load_dotenv
 import httpx
 from openai import APIConnectionError, NotFoundError, OpenAI
 from openai.types.chat import ChatCompletion, ChatCompletionChunk
 from openai.types.chat.chat_completion import Choice
 from openai.types.chat.chat_completion_chunk import ChoiceDelta
-from pydantic import BaseModel
+from pydantic import BaseModel, Field, PrivateAttr, model_validator
+from typing_extensions import Self
 
 from crewai.events.types.llm_events import LLMCallType
-from crewai.llms.base_llm import BaseLLM
-from crewai.llms.hooks.transport import HTTPTransport
+from crewai.llm.base_llm import BaseLLM
+from crewai.llm.core import CONTEXT_WINDOW_USAGE_RATIO, LLM_CONTEXT_WINDOW_SIZES
+from crewai.llm.hooks.transport import HTTPTransport
+from crewai.llm.providers.utils.common import safe_tool_conversion
 from crewai.utilities.agent_utils import is_context_length_exceeded
 from crewai.utilities.exceptions.context_window_exceeding_exception import (
     LLMContextLengthExceededError,
@@ -25,11 +29,13 @@ from crewai.utilities.types import LLMMessage
 
 if TYPE_CHECKING:
     from crewai.agent.core import Agent
-    from crewai.llms.hooks.base import BaseInterceptor
     from crewai.task import Task
     from crewai.tools.base_tool import BaseTool
 
 
+load_dotenv()
+
+
 class OpenAICompletion(BaseLLM):
     """OpenAI native completion implementation.
 
@@ -37,60 +43,56 @@ class OpenAICompletion(BaseLLM):
     offering native structured outputs, function calling, and streaming support.
     """
 
-    def __init__(
-        self,
-        model: str = "gpt-4o",
-        api_key: str | None = None,
-        base_url: str | None = None,
-        organization: str | None = None,
-        project: str | None = None,
-        timeout: float | None = None,
-        max_retries: int = 2,
-        default_headers: dict[str, str] | None = None,
-        default_query: dict[str, Any] | None = None,
-        client_params: dict[str, Any] | None = None,
-        temperature: float | None = None,
-        top_p: float | None = None,
-        frequency_penalty: float | None = None,
-        presence_penalty: float | None = None,
-        max_tokens: int | None = None,
-        max_completion_tokens: int | None = None,
-        seed: int | None = None,
-        stream: bool = False,
-        response_format: dict[str, Any] | type[BaseModel] | None = None,
-        logprobs: bool | None = None,
-        top_logprobs: int | None = None,
-        reasoning_effort: str | None = None,
-        provider: str | None = None,
-        interceptor: BaseInterceptor[httpx.Request, httpx.Response] | None = None,
-        **kwargs: Any,
-    ) -> None:
-        """Initialize OpenAI chat completion client."""
+    # Client configuration fields
+    organization: str | None = Field(default=None, description="OpenAI organization ID")
+    project: str | None = Field(default=None, description="OpenAI project ID")
+    max_retries: int = Field(default=2, description="Maximum number of retries")
+    default_headers: dict[str, str] = Field(
+        default_factory=dict, description="Default headers for requests"
+    )
+    default_query: dict[str, Any] = Field(
+        default_factory=dict, description="Default query parameters"
+    )
+    client_params: dict[str, Any] = Field(
+        default_factory=dict, description="Additional client parameters"
+    )
+    timeout: float | None = Field(default=None, description="Request timeout")
+    api_base: str | None = Field(
+        default=None, description="API base URL", deprecated=True
+    )
 
-        if provider is None:
-            provider = kwargs.pop("provider", "openai")
+    # Completion parameters
+    top_p: float | None = Field(default=None, description="Top-p sampling parameter")
+    frequency_penalty: float | None = Field(
+        default=None, description="Frequency penalty"
+    )
+    presence_penalty: float | None = Field(default=None, description="Presence penalty")
+    max_tokens: int | None = Field(default=None, description="Maximum tokens")
+    max_completion_tokens: int | None = Field(
+        None, description="Maximum completion tokens"
+    )
+    seed: int | None = Field(default=None, description="Random seed")
+    stream: bool = Field(default=False, description="Enable streaming")
+    response_format: dict[str, Any] | type[BaseModel] | None = Field(
+        default=None, description="Response format"
+    )
+    logprobs: bool | None = Field(default=None, description="Return log probabilities")
+    top_logprobs: int | None = Field(
+        default=None, description="Number of top log probabilities"
+    )
+    reasoning_effort: str | None = Field(
+        default=None, description="Reasoning effort level"
+    )
 
-        self.interceptor = interceptor
-        # Client configuration attributes
-        self.organization = organization
-        self.project = project
-        self.max_retries = max_retries
-        self.default_headers = default_headers
-        self.default_query = default_query
-        self.client_params = client_params
-        self.timeout = timeout
-        self.base_url = base_url
-        self.api_base = kwargs.pop("api_base", None)
+    _client: OpenAI = PrivateAttr(default=None)  # type: ignore[assignment]
+    is_o1_model: bool = Field(default=False, description="Whether this is an O1 model")
+    is_gpt4_model: bool = Field(
+        default=False, description="Whether this is a GPT-4 model"
+    )
 
-        super().__init__(
-            model=model,
-            temperature=temperature,
-            api_key=api_key or os.getenv("OPENAI_API_KEY"),
-            base_url=base_url,
-            timeout=timeout,
-            provider=provider,
-            **kwargs,
-        )
+    @model_validator(mode="after")
+    def setup_client(self) -> Self:
+        """Initialize OpenAI client after model validation."""
 
         client_config = self._get_client_params()
         if self.interceptor:
@@ -98,31 +100,15 @@ class OpenAICompletion(BaseLLM):
             http_client = httpx.Client(transport=transport)
             client_config["http_client"] = http_client
 
-        self.client = OpenAI(**client_config)
+        self._client = OpenAI(**client_config)
 
-        # Completion parameters
-        self.top_p = top_p
-        self.frequency_penalty = frequency_penalty
-        self.presence_penalty = presence_penalty
-        self.max_tokens = max_tokens
-        self.max_completion_tokens = max_completion_tokens
-        self.seed = seed
-        self.stream = stream
-        self.response_format = response_format
-        self.logprobs = logprobs
-        self.top_logprobs = top_logprobs
-        self.reasoning_effort = reasoning_effort
-        self.is_o1_model = "o1" in model.lower()
-        self.is_gpt4_model = "gpt-4" in model.lower()
+        self.is_o1_model = "o1" in self.model.lower()
+        self.is_gpt4_model = "gpt-4" in self.model.lower()
+
+        return self
 
     def _get_client_params(self) -> dict[str, Any]:
         """Get OpenAI client parameters."""
-
-        if self.api_key is None:
-            self.api_key = os.getenv("OPENAI_API_KEY")
-            if self.api_key is None:
-                raise ValueError("OPENAI_API_KEY is required")
-
         base_params = {
             "api_key": self.api_key,
             "organization": self.organization,
@@ -268,7 +254,6 @@ class OpenAICompletion(BaseLLM):
         self, tools: list[dict[str, BaseTool]]
     ) -> list[dict[str, Any]]:
         """Convert CrewAI tool format to OpenAI function calling format."""
-        from crewai.llms.providers.utils.common import safe_tool_conversion
 
         openai_tools = []
 
@@ -296,14 +281,14 @@ class OpenAICompletion(BaseLLM):
         self,
         params: dict[str, Any],
         available_functions: dict[str, Any] | None = None,
-        from_task: Any | None = None,
-        from_agent: Any | None = None,
+        from_task: Task | None = None,
+        from_agent: Agent | None = None,
         response_model: type[BaseModel] | None = None,
     ) -> str | Any:
         """Handle non-streaming chat completion."""
         try:
             if response_model:
-                parsed_response = self.client.beta.chat.completions.parse(
+                parsed_response = self._client.beta.chat.completions.parse(
                     **params,
                     response_format=response_model,
                 )
@@ -327,7 +312,7 @@ class OpenAICompletion(BaseLLM):
                     )
                     return structured_json
 
-            response: ChatCompletion = self.client.chat.completions.create(**params)
+            response: ChatCompletion = self._client.chat.completions.create(**params)
 
             usage = self._extract_openai_token_usage(response)
 
@@ -419,8 +404,8 @@ class OpenAICompletion(BaseLLM):
         self,
         params: dict[str, Any],
         available_functions: dict[str, Any] | None = None,
-        from_task: Any | None = None,
-        from_agent: Any | None = None,
+        from_task: Task | None = None,
+        from_agent: Agent | None = None,
         response_model: type[BaseModel] | None = None,
     ) -> str:
         """Handle streaming chat completion."""
@@ -429,7 +414,7 @@ class OpenAICompletion(BaseLLM):
 
         if response_model:
             completion_stream: Iterator[ChatCompletionChunk] = (
-                self.client.chat.completions.create(**params)
+                self._client.chat.completions.create(**params)
             )
 
             accumulated_content = ""
@@ -472,7 +457,7 @@ class OpenAICompletion(BaseLLM):
                 )
                 return accumulated_content
 
-        stream: Iterator[ChatCompletionChunk] = self.client.chat.completions.create(
+        stream: Iterator[ChatCompletionChunk] = self._client.chat.completions.create(
             **params
         )
 
@@ -560,7 +545,6 @@ class OpenAICompletion(BaseLLM):
 
     def get_context_window_size(self) -> int:
         """Get the context window size for the model."""
-        from crewai.llm import CONTEXT_WINDOW_USAGE_RATIO, LLM_CONTEXT_WINDOW_SIZES
 
         min_context = 1024
         max_context = 2097152

lib/crewai/src/crewai/llms/__init__.py

@@ -1 +1,38 @@
-"""LLM implementations for crewAI."""
+"""LLM implementations for crewAI.
+
+.. deprecated:: 1.4.0
+    The `crewai.llms` package is deprecated. Use `crewai.llm` instead.
+
+    This package was reorganized from `crewai.llms.*` to `crewai.llm.*`.
+    All submodules are redirected to their new locations in `crewai.llm.*`.
+
+    Migration guide:
+        Old: from crewai.llms.base_llm import BaseLLM
+        New: from crewai.llm.base_llm import BaseLLM
+
+        Old: from crewai.llms.hooks.base import BaseInterceptor
+        New: from crewai.llm.hooks.base import BaseInterceptor
+
+        Old: from crewai.llms.constants import OPENAI_MODELS
+        New: from crewai.llm.constants import OPENAI_MODELS
+
+        Or use top-level imports:
+        from crewai import LLM, BaseLLM
+"""
+
+import warnings
+
+from crewai.llm import LLM
+from crewai.llm.base_llm import BaseLLM
+
+
+# Issue deprecation warning when this module is imported
+warnings.warn(
+    "The 'crewai.llms' package is deprecated and will be removed in a future version. "
+    "Please use 'crewai.llm' (singular) instead. "
+    "All submodules have been reorganized from 'crewai.llms.*' to 'crewai.llm.*'.",
+    DeprecationWarning,
+    stacklevel=2,
+)
+
+__all__ = ["LLM", "BaseLLM"]

lib/crewai/src/crewai/llms/base_llm.py

@@ -1,550 +1,15 @@
-"""Base LLM abstract class for CrewAI.
+"""Deprecated: Use crewai.llm.base_llm instead.
 
-This module provides the abstract base class for all LLM implementations
-in CrewAI, including common functionality for native SDK implementations.
+.. deprecated:: 1.4.0
 """
 
-from __future__ import annotations
+import warnings
 
-from abc import ABC, abstractmethod
-from datetime import datetime
-import json
-import logging
-import re
-from typing import TYPE_CHECKING, Any, Final
 
-from pydantic import BaseModel
-
-from crewai.events.event_bus import crewai_event_bus
-from crewai.events.types.llm_events import (
-    LLMCallCompletedEvent,
-    LLMCallFailedEvent,
-    LLMCallStartedEvent,
-    LLMCallType,
-    LLMStreamChunkEvent,
+warnings.warn(
+    "crewai.llms.base_llm is deprecated. Use crewai.llm.base_llm instead.",
+    DeprecationWarning,
+    stacklevel=2,
 )
-from crewai.events.types.tool_usage_events import (
-    ToolUsageErrorEvent,
-    ToolUsageFinishedEvent,
-    ToolUsageStartedEvent,
-)
-from crewai.types.usage_metrics import UsageMetrics
 
-
-if TYPE_CHECKING:
-    from crewai.agent.core import Agent
-    from crewai.task import Task
-    from crewai.tools.base_tool import BaseTool
-    from crewai.utilities.types import LLMMessage
-
-
-DEFAULT_CONTEXT_WINDOW_SIZE: Final[int] = 4096
-DEFAULT_SUPPORTS_STOP_WORDS: Final[bool] = True
-_JSON_EXTRACTION_PATTERN: Final[re.Pattern[str]] = re.compile(r"\{.*}", re.DOTALL)
-
-
-class BaseLLM(ABC):
-    """Abstract base class for LLM implementations.
-
-    This class defines the interface that all LLM implementations must follow.
-    Users can extend this class to create custom LLM implementations that don't
-    rely on litellm's authentication mechanism.
-
-    Custom LLM implementations should handle error cases gracefully, including
-    timeouts, authentication failures, and malformed responses. They should also
-    implement proper validation for input parameters and provide clear error
-    messages when things go wrong.
-
-    Attributes:
-        model: The model identifier/name.
-        temperature: Optional temperature setting for response generation.
-        stop: A list of stop sequences that the LLM should use to stop generation.
-        additional_params: Additional provider-specific parameters.
-    """
-
-    is_litellm: bool = False
-
-    def __init__(
-        self,
-        model: str,
-        temperature: float | None = None,
-        api_key: str | None = None,
-        base_url: str | None = None,
-        provider: str | None = None,
-        **kwargs: Any,
-    ) -> None:
-        """Initialize the BaseLLM with default attributes.
-
-        Args:
-            model: The model identifier/name.
-            temperature: Optional temperature setting for response generation.
-            stop: Optional list of stop sequences for generation.
-            **kwargs: Additional provider-specific parameters.
-        """
-        if not model:
-            raise ValueError("Model name is required and cannot be empty")
-
-        self.model = model
-        self.temperature = temperature
-        self.api_key = api_key
-        self.base_url = base_url
-        # Store additional parameters for provider-specific use
-        self.additional_params = kwargs
-        self._provider = provider or "openai"
-
-        stop = kwargs.pop("stop", None)
-        if stop is None:
-            self.stop: list[str] = []
-        elif isinstance(stop, str):
-            self.stop = [stop]
-        elif isinstance(stop, list):
-            self.stop = stop
-        else:
-            self.stop = []
-
-        self._token_usage = {
-            "total_tokens": 0,
-            "prompt_tokens": 0,
-            "completion_tokens": 0,
-            "successful_requests": 0,
-            "cached_prompt_tokens": 0,
-        }
-
-    @property
-    def provider(self) -> str:
-        """Get the provider of the LLM."""
-        return self._provider
-
-    @provider.setter
-    def provider(self, value: str) -> None:
-        """Set the provider of the LLM."""
-        self._provider = value
-
-    @abstractmethod
-    def call(
-        self,
-        messages: str | list[LLMMessage],
-        tools: list[dict[str, BaseTool]] | None = None,
-        callbacks: list[Any] | None = None,
-        available_functions: dict[str, Any] | None = None,
-        from_task: Task | None = None,
-        from_agent: Agent | None = None,
-        response_model: type[BaseModel] | None = None,
-    ) -> str | Any:
-        """Call the LLM with the given messages.
-
-        Args:
-            messages: Input messages for the LLM.
-                     Can be a string or list of message dictionaries.
-                     If string, it will be converted to a single user message.
-                     If list, each dict must have 'role' and 'content' keys.
-            tools: Optional list of tool schemas for function calling.
-                  Each tool should define its name, description, and parameters.
-            callbacks: Optional list of callback functions to be executed
-                      during and after the LLM call.
-            available_functions: Optional dict mapping function names to callables
-                               that can be invoked by the LLM.
-            from_task: Optional task caller to be used for the LLM call.
-            from_agent: Optional agent caller to be used for the LLM call.
-            response_model: Optional response model to be used for the LLM call.
-
-        Returns:
-            Either a text response from the LLM (str) or
-            the result of a tool function call (Any).
-
-        Raises:
-            ValueError: If the messages format is invalid.
-            TimeoutError: If the LLM request times out.
-            RuntimeError: If the LLM request fails for other reasons.
-        """
-
-    def _convert_tools_for_interference(
-        self, tools: list[dict[str, BaseTool]]
-    ) -> list[dict[str, BaseTool]]:
-        """Convert tools to a format that can be used for interference.
-
-        Args:
-            tools: List of tools to convert.
-
-        Returns:
-            List of converted tools (default implementation returns as-is)
-        """
-        return tools
-
-    def supports_stop_words(self) -> bool:
-        """Check if the LLM supports stop words.
-
-        Returns:
-            True if the LLM supports stop words, False otherwise.
-        """
-        return DEFAULT_SUPPORTS_STOP_WORDS
-
-    def _supports_stop_words_implementation(self) -> bool:
-        """Check if stop words are configured for this LLM instance.
-
-        Native providers can override supports_stop_words() to return this value
-        to ensure consistent behavior based on whether stop words are actually configured.
-
-        Returns:
-            True if stop words are configured and can be applied
-        """
-        return bool(self.stop)
-
-    def _apply_stop_words(self, content: str) -> str:
-        """Apply stop words to truncate response content.
-
-        This method provides consistent stop word behavior across all native SDK providers.
-        Native providers should call this method to post-process their responses.
-
-        Args:
-            content: The raw response content from the LLM
-
-        Returns:
-            Content truncated at the first occurrence of any stop word
-
-        Example:
-            >>> llm = MyNativeLLM(stop=["Observation:", "Final Answer:"])
-            >>> response = (
-            ...     "I need to search.\\n\\nAction: search\\nObservation: Found results"
-            ... )
-            >>> llm._apply_stop_words(response)
-            "I need to search.\\n\\nAction: search"
-        """
-        if not self.stop or not content:
-            return content
-
-        # Find the earliest occurrence of any stop word
-        earliest_stop_pos = len(content)
-        found_stop_word = None
-
-        for stop_word in self.stop:
-            stop_pos = content.find(stop_word)
-            if stop_pos != -1 and stop_pos < earliest_stop_pos:
-                earliest_stop_pos = stop_pos
-                found_stop_word = stop_word
-
-        # Truncate at the stop word if found
-        if found_stop_word is not None:
-            truncated = content[:earliest_stop_pos].strip()
-            logging.debug(
-                f"Applied stop word '{found_stop_word}' at position {earliest_stop_pos}"
-            )
-            return truncated
-
-        return content
-
-    def get_context_window_size(self) -> int:
-        """Get the context window size for the LLM.
-
-        Returns:
-            The number of tokens/characters the model can handle.
-        """
-        # Default implementation - subclasses should override with model-specific values
-        return DEFAULT_CONTEXT_WINDOW_SIZE
-
-    # Common helper methods for native SDK implementations
-
-    def _emit_call_started_event(
-        self,
-        messages: str | list[LLMMessage],
-        tools: list[dict[str, BaseTool]] | None = None,
-        callbacks: list[Any] | None = None,
-        available_functions: dict[str, Any] | None = None,
-        from_task: Task | None = None,
-        from_agent: Agent | None = None,
-    ) -> None:
-        """Emit LLM call started event."""
-        if not hasattr(crewai_event_bus, "emit"):
-            raise ValueError("crewai_event_bus does not have an emit method") from None
-
-        crewai_event_bus.emit(
-            self,
-            event=LLMCallStartedEvent(
-                messages=messages,
-                tools=tools,
-                callbacks=callbacks,
-                available_functions=available_functions,
-                from_task=from_task,
-                from_agent=from_agent,
-                model=self.model,
-            ),
-        )
-
-    def _emit_call_completed_event(
-        self,
-        response: Any,
-        call_type: LLMCallType,
-        from_task: Task | None = None,
-        from_agent: Agent | None = None,
-        messages: str | list[dict[str, Any]] | None = None,
-    ) -> None:
-        """Emit LLM call completed event."""
-        crewai_event_bus.emit(
-            self,
-            event=LLMCallCompletedEvent(
-                messages=messages,
-                response=response,
-                call_type=call_type,
-                from_task=from_task,
-                from_agent=from_agent,
-                model=self.model,
-            ),
-        )
-
-    def _emit_call_failed_event(
-        self,
-        error: str,
-        from_task: Task | None = None,
-        from_agent: Agent | None = None,
-    ) -> None:
-        """Emit LLM call failed event."""
-        if not hasattr(crewai_event_bus, "emit"):
-            raise ValueError("crewai_event_bus does not have an emit method") from None
-
-        crewai_event_bus.emit(
-            self,
-            event=LLMCallFailedEvent(
-                error=error,
-                from_task=from_task,
-                from_agent=from_agent,
-            ),
-        )
-
-    def _emit_stream_chunk_event(
-        self,
-        chunk: str,
-        from_task: Task | None = None,
-        from_agent: Agent | None = None,
-        tool_call: dict[str, Any] | None = None,
-    ) -> None:
-        """Emit stream chunk event."""
-        if not hasattr(crewai_event_bus, "emit"):
-            raise ValueError("crewai_event_bus does not have an emit method") from None
-
-        crewai_event_bus.emit(
-            self,
-            event=LLMStreamChunkEvent(
-                chunk=chunk,
-                tool_call=tool_call,
-                from_task=from_task,
-                from_agent=from_agent,
-            ),
-        )
-
-    def _handle_tool_execution(
-        self,
-        function_name: str,
-        function_args: dict[str, Any],
-        available_functions: dict[str, Any],
-        from_task: Task | None = None,
-        from_agent: Agent | None = None,
-    ) -> str | None:
-        """Handle tool execution with proper event emission.
-
-        Args:
-            function_name: Name of the function to execute
-            function_args: Arguments to pass to the function
-            available_functions: Dict of available functions
-            from_task: Optional task object
-            from_agent: Optional agent object
-
-        Returns:
-            Result of function execution or None if function not found
-        """
-        if function_name not in available_functions:
-            logging.warning(
-                f"Function '{function_name}' not found in available functions"
-            )
-            return None
-
-        try:
-            # Emit tool usage started event
-            started_at = datetime.now()
-
-            crewai_event_bus.emit(
-                self,
-                event=ToolUsageStartedEvent(
-                    tool_name=function_name,
-                    tool_args=function_args,
-                    from_agent=from_agent,
-                    from_task=from_task,
-                ),
-            )
-
-            # Execute the function
-            fn = available_functions[function_name]
-            result = fn(**function_args)
-
-            # Emit tool usage finished event
-            crewai_event_bus.emit(
-                self,
-                event=ToolUsageFinishedEvent(
-                    output=result,
-                    tool_name=function_name,
-                    tool_args=function_args,
-                    started_at=started_at,
-                    finished_at=datetime.now(),
-                    from_task=from_task,
-                    from_agent=from_agent,
-                ),
-            )
-
-            # Emit LLM call completed event for tool call
-            self._emit_call_completed_event(
-                response=result,
-                call_type=LLMCallType.TOOL_CALL,
-                from_task=from_task,
-                from_agent=from_agent,
-            )
-
-            return str(result)
-
-        except Exception as e:
-            error_msg = f"Error executing function '{function_name}': {e!s}"
-            logging.error(error_msg)
-
-            # Emit tool usage error event
-            if not hasattr(crewai_event_bus, "emit"):
-                raise ValueError(
-                    "crewai_event_bus does not have an emit method"
-                ) from None
-
-            crewai_event_bus.emit(
-                self,
-                event=ToolUsageErrorEvent(
-                    tool_name=function_name,
-                    tool_args=function_args,
-                    error=error_msg,
-                    from_task=from_task,
-                    from_agent=from_agent,
-                ),
-            )
-
-            # Emit LLM call failed event
-            self._emit_call_failed_event(
-                error=error_msg,
-                from_task=from_task,
-                from_agent=from_agent,
-            )
-
-            return None
-
-    def _format_messages(self, messages: str | list[LLMMessage]) -> list[LLMMessage]:
-        """Convert messages to standard format.
-
-        Args:
-            messages: Input messages (string or list of message dicts)
-
-        Returns:
-            List of message dictionaries with 'role' and 'content' keys
-
-        Raises:
-            ValueError: If message format is invalid
-        """
-        if isinstance(messages, str):
-            return [{"role": "user", "content": messages}]
-
-        # Validate message format
-        for i, msg in enumerate(messages):
-            if not isinstance(msg, dict):
-                raise ValueError(f"Message at index {i} must be a dictionary")
-            if "role" not in msg or "content" not in msg:
-                raise ValueError(
-                    f"Message at index {i} must have 'role' and 'content' keys"
-                )
-
-        return messages
-
-    @staticmethod
-    def _validate_structured_output(
-        response: str,
-        response_format: type[BaseModel] | None,
-    ) -> str | BaseModel:
-        """Validate and parse structured output.
-
-        Args:
-            response: Raw response string
-            response_format: Optional Pydantic model for structured output
-
-        Returns:
-            Parsed response (BaseModel instance if response_format provided, otherwise string)
-
-        Raises:
-            ValueError: If structured output validation fails
-        """
-        if response_format is None:
-            return response
-
-        try:
-            # Try to parse as JSON first
-            if response.strip().startswith("{") or response.strip().startswith("["):
-                data = json.loads(response)
-                return response_format.model_validate(data)
-
-            json_match = _JSON_EXTRACTION_PATTERN.search(response)
-            if json_match:
-                data = json.loads(json_match.group())
-                return response_format.model_validate(data)
-
-            raise ValueError("No JSON found in response")
-
-        except (json.JSONDecodeError, ValueError) as e:
-            logging.warning(f"Failed to parse structured output: {e}")
-            raise ValueError(
-                f"Failed to parse response into {response_format.__name__}: {e}"
-            ) from e
-
-    @staticmethod
-    def _extract_provider(model: str) -> str:
-        """Extract provider from model string.
-
-        Args:
-            model: Model string (e.g., 'openai/gpt-4' or 'gpt-4')
-
-        Returns:
-            Provider name (e.g., 'openai')
-        """
-        if "/" in model:
-            return model.partition("/")[0]
-        return "openai"  # Default provider
-
-    def _track_token_usage_internal(self, usage_data: dict[str, Any]) -> None:
-        """Track token usage internally in the LLM instance.
-
-        Args:
-            usage_data: Token usage data from the API response
-        """
-        # Extract tokens in a provider-agnostic way
-        prompt_tokens = (
-            usage_data.get("prompt_tokens")
-            or usage_data.get("prompt_token_count")
-            or usage_data.get("input_tokens")
-            or 0
-        )
-
-        completion_tokens = (
-            usage_data.get("completion_tokens")
-            or usage_data.get("candidates_token_count")
-            or usage_data.get("output_tokens")
-            or 0
-        )
-
-        cached_tokens = (
-            usage_data.get("cached_tokens")
-            or usage_data.get("cached_prompt_tokens")
-            or 0
-        )
-
-        self._token_usage["prompt_tokens"] += prompt_tokens
-        self._token_usage["completion_tokens"] += completion_tokens
-        self._token_usage["total_tokens"] += prompt_tokens + completion_tokens
-        self._token_usage["successful_requests"] += 1
-        self._token_usage["cached_prompt_tokens"] += cached_tokens
-
-    def get_token_usage_summary(self) -> UsageMetrics:
-        """Get summary of token usage for this LLM instance.
-
-        Returns:
-            Dictionary with token usage totals
-        """
-        return UsageMetrics(**self._token_usage)
+from crewai.llm.base_llm import *  # noqa: E402, F403

lib/crewai/src/crewai/llms/constants.py

@@ -1,558 +1,15 @@
-from typing import Literal, TypeAlias
+"""Deprecated: Use crewai.llm.constants instead.
+
+.. deprecated:: 1.4.0
+"""
+
+import warnings
 
 
-OpenAIModels: TypeAlias = Literal[
-    "gpt-3.5-turbo",
-    "gpt-3.5-turbo-0125",
-    "gpt-3.5-turbo-0301",
-    "gpt-3.5-turbo-0613",
-    "gpt-3.5-turbo-1106",
-    "gpt-3.5-turbo-16k",
-    "gpt-3.5-turbo-16k-0613",
-    "gpt-3.5-turbo-instruct",
-    "gpt-3.5-turbo-instruct-0914",
-    "gpt-4",
-    "gpt-4-0125-preview",
-    "gpt-4-0314",
-    "gpt-4-0613",
-    "gpt-4-1106-preview",
-    "gpt-4-32k",
-    "gpt-4-32k-0314",
-    "gpt-4-32k-0613",
-    "gpt-4-turbo",
-    "gpt-4-turbo-2024-04-09",
-    "gpt-4-turbo-preview",
-    "gpt-4-vision-preview",
-    "gpt-4.1",
-    "gpt-4.1-2025-04-14",
-    "gpt-4.1-mini",
-    "gpt-4.1-mini-2025-04-14",
-    "gpt-4.1-nano",
-    "gpt-4.1-nano-2025-04-14",
-    "gpt-4o",
-    "gpt-4o-2024-05-13",
-    "gpt-4o-2024-08-06",
-    "gpt-4o-2024-11-20",
-    "gpt-4o-audio-preview",
-    "gpt-4o-audio-preview-2024-10-01",
-    "gpt-4o-audio-preview-2024-12-17",
-    "gpt-4o-audio-preview-2025-06-03",
-    "gpt-4o-mini",
-    "gpt-4o-mini-2024-07-18",
-    "gpt-4o-mini-audio-preview",
-    "gpt-4o-mini-audio-preview-2024-12-17",
-    "gpt-4o-mini-realtime-preview",
-    "gpt-4o-mini-realtime-preview-2024-12-17",
-    "gpt-4o-mini-search-preview",
-    "gpt-4o-mini-search-preview-2025-03-11",
-    "gpt-4o-mini-transcribe",
-    "gpt-4o-mini-tts",
-    "gpt-4o-realtime-preview",
-    "gpt-4o-realtime-preview-2024-10-01",
-    "gpt-4o-realtime-preview-2024-12-17",
-    "gpt-4o-realtime-preview-2025-06-03",
-    "gpt-4o-search-preview",
-    "gpt-4o-search-preview-2025-03-11",
-    "gpt-4o-transcribe",
-    "gpt-4o-transcribe-diarize",
-    "gpt-5",
-    "gpt-5-2025-08-07",
-    "gpt-5-chat",
-    "gpt-5-chat-latest",
-    "gpt-5-codex",
-    "gpt-5-mini",
-    "gpt-5-mini-2025-08-07",
-    "gpt-5-nano",
-    "gpt-5-nano-2025-08-07",
-    "gpt-5-pro",
-    "gpt-5-pro-2025-10-06",
-    "gpt-5-search-api",
-    "gpt-5-search-api-2025-10-14",
-    "gpt-audio",
-    "gpt-audio-2025-08-28",
-    "gpt-audio-mini",
-    "gpt-audio-mini-2025-10-06",
-    "gpt-image-1",
-    "gpt-image-1-mini",
-    "gpt-realtime",
-    "gpt-realtime-2025-08-28",
-    "gpt-realtime-mini",
-    "gpt-realtime-mini-2025-10-06",
-    "o1",
-    "o1-preview",
-    "o1-2024-12-17",
-    "o1-mini",
-    "o1-mini-2024-09-12",
-    "o1-pro",
-    "o1-pro-2025-03-19",
-    "o3-mini",
-    "o3",
-    "o4-mini",
-    "whisper-1",
-]
-OPENAI_MODELS: list[OpenAIModels] = [
-    "gpt-3.5-turbo",
-    "gpt-3.5-turbo-0125",
-    "gpt-3.5-turbo-0301",
-    "gpt-3.5-turbo-0613",
-    "gpt-3.5-turbo-1106",
-    "gpt-3.5-turbo-16k",
-    "gpt-3.5-turbo-16k-0613",
-    "gpt-3.5-turbo-instruct",
-    "gpt-3.5-turbo-instruct-0914",
-    "gpt-4",
-    "gpt-4-0125-preview",
-    "gpt-4-0314",
-    "gpt-4-0613",
-    "gpt-4-1106-preview",
-    "gpt-4-32k",
-    "gpt-4-32k-0314",
-    "gpt-4-32k-0613",
-    "gpt-4-turbo",
-    "gpt-4-turbo-2024-04-09",
-    "gpt-4-turbo-preview",
-    "gpt-4-vision-preview",
-    "gpt-4.1",
-    "gpt-4.1-2025-04-14",
-    "gpt-4.1-mini",
-    "gpt-4.1-mini-2025-04-14",
-    "gpt-4.1-nano",
-    "gpt-4.1-nano-2025-04-14",
-    "gpt-4o",
-    "gpt-4o-2024-05-13",
-    "gpt-4o-2024-08-06",
-    "gpt-4o-2024-11-20",
-    "gpt-4o-audio-preview",
-    "gpt-4o-audio-preview-2024-10-01",
-    "gpt-4o-audio-preview-2024-12-17",
-    "gpt-4o-audio-preview-2025-06-03",
-    "gpt-4o-mini",
-    "gpt-4o-mini-2024-07-18",
-    "gpt-4o-mini-audio-preview",
-    "gpt-4o-mini-audio-preview-2024-12-17",
-    "gpt-4o-mini-realtime-preview",
-    "gpt-4o-mini-realtime-preview-2024-12-17",
-    "gpt-4o-mini-search-preview",
-    "gpt-4o-mini-search-preview-2025-03-11",
-    "gpt-4o-mini-transcribe",
-    "gpt-4o-mini-tts",
-    "gpt-4o-realtime-preview",
-    "gpt-4o-realtime-preview-2024-10-01",
-    "gpt-4o-realtime-preview-2024-12-17",
-    "gpt-4o-realtime-preview-2025-06-03",
-    "gpt-4o-search-preview",
-    "gpt-4o-search-preview-2025-03-11",
-    "gpt-4o-transcribe",
-    "gpt-4o-transcribe-diarize",
-    "gpt-5",
-    "gpt-5-2025-08-07",
-    "gpt-5-chat",
-    "gpt-5-chat-latest",
-    "gpt-5-codex",
-    "gpt-5-mini",
-    "gpt-5-mini-2025-08-07",
-    "gpt-5-nano",
-    "gpt-5-nano-2025-08-07",
-    "gpt-5-pro",
-    "gpt-5-pro-2025-10-06",
-    "gpt-5-search-api",
-    "gpt-5-search-api-2025-10-14",
-    "gpt-audio",
-    "gpt-audio-2025-08-28",
-    "gpt-audio-mini",
-    "gpt-audio-mini-2025-10-06",
-    "gpt-image-1",
-    "gpt-image-1-mini",
-    "gpt-realtime",
-    "gpt-realtime-2025-08-28",
-    "gpt-realtime-mini",
-    "gpt-realtime-mini-2025-10-06",
-    "o1",
-    "o1-preview",
-    "o1-2024-12-17",
-    "o1-mini",
-    "o1-mini-2024-09-12",
-    "o1-pro",
-    "o1-pro-2025-03-19",
-    "o3-mini",
-    "o3",
-    "o4-mini",
-    "whisper-1",
-]
+warnings.warn(
+    "crewai.llms.constants is deprecated. Use crewai.llm.constants instead.",
+    DeprecationWarning,
+    stacklevel=2,
+)
 
-
-AnthropicModels: TypeAlias = Literal[
-    "claude-3-7-sonnet-latest",
-    "claude-3-7-sonnet-20250219",
-    "claude-3-5-haiku-latest",
-    "claude-3-5-haiku-20241022",
-    "claude-haiku-4-5",
-    "claude-haiku-4-5-20251001",
-    "claude-sonnet-4-20250514",
-    "claude-sonnet-4-0",
-    "claude-4-sonnet-20250514",
-    "claude-sonnet-4-5",
-    "claude-sonnet-4-5-20250929",
-    "claude-3-5-sonnet-latest",
-    "claude-3-5-sonnet-20241022",
-    "claude-3-5-sonnet-20240620",
-    "claude-opus-4-0",
-    "claude-opus-4-20250514",
-    "claude-4-opus-20250514",
-    "claude-opus-4-1",
-    "claude-opus-4-1-20250805",
-    "claude-3-opus-latest",
-    "claude-3-opus-20240229",
-    "claude-3-sonnet-20240229",
-    "claude-3-haiku-latest",
-    "claude-3-haiku-20240307",
-]
-ANTHROPIC_MODELS: list[AnthropicModels] = [
-    "claude-3-7-sonnet-latest",
-    "claude-3-7-sonnet-20250219",
-    "claude-3-5-haiku-latest",
-    "claude-3-5-haiku-20241022",
-    "claude-haiku-4-5",
-    "claude-haiku-4-5-20251001",
-    "claude-sonnet-4-20250514",
-    "claude-sonnet-4-0",
-    "claude-4-sonnet-20250514",
-    "claude-sonnet-4-5",
-    "claude-sonnet-4-5-20250929",
-    "claude-3-5-sonnet-latest",
-    "claude-3-5-sonnet-20241022",
-    "claude-3-5-sonnet-20240620",
-    "claude-opus-4-0",
-    "claude-opus-4-20250514",
-    "claude-4-opus-20250514",
-    "claude-opus-4-1",
-    "claude-opus-4-1-20250805",
-    "claude-3-opus-latest",
-    "claude-3-opus-20240229",
-    "claude-3-sonnet-20240229",
-    "claude-3-haiku-latest",
-    "claude-3-haiku-20240307",
-]
-
-GeminiModels: TypeAlias = Literal[
-    "gemini-2.5-pro",
-    "gemini-2.5-pro-preview-03-25",
-    "gemini-2.5-pro-preview-05-06",
-    "gemini-2.5-pro-preview-06-05",
-    "gemini-2.5-flash",
-    "gemini-2.5-flash-preview-05-20",
-    "gemini-2.5-flash-preview-04-17",
-    "gemini-2.5-flash-image",
-    "gemini-2.5-flash-image-preview",
-    "gemini-2.5-flash-lite",
-    "gemini-2.5-flash-lite-preview-06-17",
-    "gemini-2.5-flash-preview-09-2025",
-    "gemini-2.5-flash-lite-preview-09-2025",
-    "gemini-2.5-flash-preview-tts",
-    "gemini-2.5-pro-preview-tts",
-    "gemini-2.5-computer-use-preview-10-2025",
-    "gemini-2.0-flash",
-    "gemini-2.0-flash-001",
-    "gemini-2.0-flash-exp",
-    "gemini-2.0-flash-exp-image-generation",
-    "gemini-2.0-flash-lite",
-    "gemini-2.0-flash-lite-001",
-    "gemini-2.0-flash-lite-preview",
-    "gemini-2.0-flash-lite-preview-02-05",
-    "gemini-2.0-flash-preview-image-generation",
-    "gemini-2.0-flash-thinking-exp",
-    "gemini-2.0-flash-thinking-exp-01-21",
-    "gemini-2.0-flash-thinking-exp-1219",
-    "gemini-2.0-pro-exp",
-    "gemini-2.0-pro-exp-02-05",
-    "gemini-exp-1206",
-    "gemini-1.5-pro",
-    "gemini-1.5-flash",
-    "gemini-1.5-flash-8b",
-    "gemini-flash-latest",
-    "gemini-flash-lite-latest",
-    "gemini-pro-latest",
-    "gemini-2.0-flash-live-001",
-    "gemini-live-2.5-flash-preview",
-    "gemini-2.5-flash-live-preview",
-    "gemini-robotics-er-1.5-preview",
-    "gemini-gemma-2-27b-it",
-    "gemini-gemma-2-9b-it",
-    "gemma-3-1b-it",
-    "gemma-3-4b-it",
-    "gemma-3-12b-it",
-    "gemma-3-27b-it",
-    "gemma-3n-e2b-it",
-    "gemma-3n-e4b-it",
-    "learnlm-2.0-flash-experimental",
-]
-GEMINI_MODELS: list[GeminiModels] = [
-    "gemini-2.5-pro",
-    "gemini-2.5-pro-preview-03-25",
-    "gemini-2.5-pro-preview-05-06",
-    "gemini-2.5-pro-preview-06-05",
-    "gemini-2.5-flash",
-    "gemini-2.5-flash-preview-05-20",
-    "gemini-2.5-flash-preview-04-17",
-    "gemini-2.5-flash-image",
-    "gemini-2.5-flash-image-preview",
-    "gemini-2.5-flash-lite",
-    "gemini-2.5-flash-lite-preview-06-17",
-    "gemini-2.5-flash-preview-09-2025",
-    "gemini-2.5-flash-lite-preview-09-2025",
-    "gemini-2.5-flash-preview-tts",
-    "gemini-2.5-pro-preview-tts",
-    "gemini-2.5-computer-use-preview-10-2025",
-    "gemini-2.0-flash",
-    "gemini-2.0-flash-001",
-    "gemini-2.0-flash-exp",
-    "gemini-2.0-flash-exp-image-generation",
-    "gemini-2.0-flash-lite",
-    "gemini-2.0-flash-lite-001",
-    "gemini-2.0-flash-lite-preview",
-    "gemini-2.0-flash-lite-preview-02-05",
-    "gemini-2.0-flash-preview-image-generation",
-    "gemini-2.0-flash-thinking-exp",
-    "gemini-2.0-flash-thinking-exp-01-21",
-    "gemini-2.0-flash-thinking-exp-1219",
-    "gemini-2.0-pro-exp",
-    "gemini-2.0-pro-exp-02-05",
-    "gemini-exp-1206",
-    "gemini-1.5-pro",
-    "gemini-1.5-flash",
-    "gemini-1.5-flash-8b",
-    "gemini-flash-latest",
-    "gemini-flash-lite-latest",
-    "gemini-pro-latest",
-    "gemini-2.0-flash-live-001",
-    "gemini-live-2.5-flash-preview",
-    "gemini-2.5-flash-live-preview",
-    "gemini-robotics-er-1.5-preview",
-    "gemini-gemma-2-27b-it",
-    "gemini-gemma-2-9b-it",
-    "gemma-3-1b-it",
-    "gemma-3-4b-it",
-    "gemma-3-12b-it",
-    "gemma-3-27b-it",
-    "gemma-3n-e2b-it",
-    "gemma-3n-e4b-it",
-    "learnlm-2.0-flash-experimental",
-]
-
-
-AzureModels: TypeAlias = Literal[
-    "gpt-3.5-turbo",
-    "gpt-3.5-turbo-0301",
-    "gpt-3.5-turbo-0613",
-    "gpt-3.5-turbo-16k",
-    "gpt-3.5-turbo-16k-0613",
-    "gpt-35-turbo",
-    "gpt-35-turbo-0125",
-    "gpt-35-turbo-1106",
-    "gpt-35-turbo-16k-0613",
-    "gpt-35-turbo-instruct-0914",
-    "gpt-4",
-    "gpt-4-0314",
-    "gpt-4-0613",
-    "gpt-4-1106-preview",
-    "gpt-4-0125-preview",
-    "gpt-4-32k",
-    "gpt-4-32k-0314",
-    "gpt-4-32k-0613",
-    "gpt-4-turbo",
-    "gpt-4-turbo-2024-04-09",
-    "gpt-4-vision",
-    "gpt-4o",
-    "gpt-4o-2024-05-13",
-    "gpt-4o-2024-08-06",
-    "gpt-4o-2024-11-20",
-    "gpt-4o-mini",
-    "gpt-5",
-    "o1",
-    "o1-mini",
-    "o1-preview",
-    "o3-mini",
-    "o3",
-    "o4-mini",
-]
-AZURE_MODELS: list[AzureModels] = [
-    "gpt-3.5-turbo",
-    "gpt-3.5-turbo-0301",
-    "gpt-3.5-turbo-0613",
-    "gpt-3.5-turbo-16k",
-    "gpt-3.5-turbo-16k-0613",
-    "gpt-35-turbo",
-    "gpt-35-turbo-0125",
-    "gpt-35-turbo-1106",
-    "gpt-35-turbo-16k-0613",
-    "gpt-35-turbo-instruct-0914",
-    "gpt-4",
-    "gpt-4-0314",
-    "gpt-4-0613",
-    "gpt-4-1106-preview",
-    "gpt-4-0125-preview",
-    "gpt-4-32k",
-    "gpt-4-32k-0314",
-    "gpt-4-32k-0613",
-    "gpt-4-turbo",
-    "gpt-4-turbo-2024-04-09",
-    "gpt-4-vision",
-    "gpt-4o",
-    "gpt-4o-2024-05-13",
-    "gpt-4o-2024-08-06",
-    "gpt-4o-2024-11-20",
-    "gpt-4o-mini",
-    "gpt-5",
-    "o1",
-    "o1-mini",
-    "o1-preview",
-    "o3-mini",
-    "o3",
-    "o4-mini",
-]
-
-
-BedrockModels: TypeAlias = Literal[
-    "ai21.jamba-1-5-large-v1:0",
-    "ai21.jamba-1-5-mini-v1:0",
-    "amazon.nova-lite-v1:0",
-    "amazon.nova-lite-v1:0:24k",
-    "amazon.nova-lite-v1:0:300k",
-    "amazon.nova-micro-v1:0",
-    "amazon.nova-micro-v1:0:128k",
-    "amazon.nova-micro-v1:0:24k",
-    "amazon.nova-premier-v1:0",
-    "amazon.nova-premier-v1:0:1000k",
-    "amazon.nova-premier-v1:0:20k",
-    "amazon.nova-premier-v1:0:8k",
-    "amazon.nova-premier-v1:0:mm",
-    "amazon.nova-pro-v1:0",
-    "amazon.nova-pro-v1:0:24k",
-    "amazon.nova-pro-v1:0:300k",
-    "amazon.titan-text-express-v1",
-    "amazon.titan-text-express-v1:0:8k",
-    "amazon.titan-text-lite-v1",
-    "amazon.titan-text-lite-v1:0:4k",
-    "amazon.titan-tg1-large",
-    "anthropic.claude-3-5-haiku-20241022-v1:0",
-    "anthropic.claude-3-5-sonnet-20240620-v1:0",
-    "anthropic.claude-3-5-sonnet-20241022-v2:0",
-    "anthropic.claude-3-7-sonnet-20250219-v1:0",
-    "anthropic.claude-3-haiku-20240307-v1:0",
-    "anthropic.claude-3-haiku-20240307-v1:0:200k",
-    "anthropic.claude-3-haiku-20240307-v1:0:48k",
-    "anthropic.claude-3-opus-20240229-v1:0",
-    "anthropic.claude-3-opus-20240229-v1:0:12k",
-    "anthropic.claude-3-opus-20240229-v1:0:200k",
-    "anthropic.claude-3-opus-20240229-v1:0:28k",
-    "anthropic.claude-3-sonnet-20240229-v1:0",
-    "anthropic.claude-3-sonnet-20240229-v1:0:200k",
-    "anthropic.claude-3-sonnet-20240229-v1:0:28k",
-    "anthropic.claude-haiku-4-5-20251001-v1:0",
-    "anthropic.claude-instant-v1:2:100k",
-    "anthropic.claude-opus-4-1-20250805-v1:0",
-    "anthropic.claude-opus-4-20250514-v1:0",
-    "anthropic.claude-sonnet-4-20250514-v1:0",
-    "anthropic.claude-sonnet-4-5-20250929-v1:0",
-    "anthropic.claude-v2:0:100k",
-    "anthropic.claude-v2:0:18k",
-    "anthropic.claude-v2:1:18k",
-    "anthropic.claude-v2:1:200k",
-    "cohere.command-r-plus-v1:0",
-    "cohere.command-r-v1:0",
-    "cohere.rerank-v3-5:0",
-    "deepseek.r1-v1:0",
-    "meta.llama3-1-70b-instruct-v1:0",
-    "meta.llama3-1-8b-instruct-v1:0",
-    "meta.llama3-2-11b-instruct-v1:0",
-    "meta.llama3-2-1b-instruct-v1:0",
-    "meta.llama3-2-3b-instruct-v1:0",
-    "meta.llama3-2-90b-instruct-v1:0",
-    "meta.llama3-3-70b-instruct-v1:0",
-    "meta.llama3-70b-instruct-v1:0",
-    "meta.llama3-8b-instruct-v1:0",
-    "meta.llama4-maverick-17b-instruct-v1:0",
-    "meta.llama4-scout-17b-instruct-v1:0",
-    "mistral.mistral-7b-instruct-v0:2",
-    "mistral.mistral-large-2402-v1:0",
-    "mistral.mistral-small-2402-v1:0",
-    "mistral.mixtral-8x7b-instruct-v0:1",
-    "mistral.pixtral-large-2502-v1:0",
-    "openai.gpt-oss-120b-1:0",
-    "openai.gpt-oss-20b-1:0",
-    "qwen.qwen3-32b-v1:0",
-    "qwen.qwen3-coder-30b-a3b-v1:0",
-    "twelvelabs.pegasus-1-2-v1:0",
-]
-BEDROCK_MODELS: list[BedrockModels] = [
-    "ai21.jamba-1-5-large-v1:0",
-    "ai21.jamba-1-5-mini-v1:0",
-    "amazon.nova-lite-v1:0",
-    "amazon.nova-lite-v1:0:24k",
-    "amazon.nova-lite-v1:0:300k",
-    "amazon.nova-micro-v1:0",
-    "amazon.nova-micro-v1:0:128k",
-    "amazon.nova-micro-v1:0:24k",
-    "amazon.nova-premier-v1:0",
-    "amazon.nova-premier-v1:0:1000k",
-    "amazon.nova-premier-v1:0:20k",
-    "amazon.nova-premier-v1:0:8k",
-    "amazon.nova-premier-v1:0:mm",
-    "amazon.nova-pro-v1:0",
-    "amazon.nova-pro-v1:0:24k",
-    "amazon.nova-pro-v1:0:300k",
-    "amazon.titan-text-express-v1",
-    "amazon.titan-text-express-v1:0:8k",
-    "amazon.titan-text-lite-v1",
-    "amazon.titan-text-lite-v1:0:4k",
-    "amazon.titan-tg1-large",
-    "anthropic.claude-3-5-haiku-20241022-v1:0",
-    "anthropic.claude-3-5-sonnet-20240620-v1:0",
-    "anthropic.claude-3-5-sonnet-20241022-v2:0",
-    "anthropic.claude-3-7-sonnet-20250219-v1:0",
-    "anthropic.claude-3-haiku-20240307-v1:0",
-    "anthropic.claude-3-haiku-20240307-v1:0:200k",
-    "anthropic.claude-3-haiku-20240307-v1:0:48k",
-    "anthropic.claude-3-opus-20240229-v1:0",
-    "anthropic.claude-3-opus-20240229-v1:0:12k",
-    "anthropic.claude-3-opus-20240229-v1:0:200k",
-    "anthropic.claude-3-opus-20240229-v1:0:28k",
-    "anthropic.claude-3-sonnet-20240229-v1:0",
-    "anthropic.claude-3-sonnet-20240229-v1:0:200k",
-    "anthropic.claude-3-sonnet-20240229-v1:0:28k",
-    "anthropic.claude-haiku-4-5-20251001-v1:0",
-    "anthropic.claude-instant-v1:2:100k",
-    "anthropic.claude-opus-4-1-20250805-v1:0",
-    "anthropic.claude-opus-4-20250514-v1:0",
-    "anthropic.claude-sonnet-4-20250514-v1:0",
-    "anthropic.claude-sonnet-4-5-20250929-v1:0",
-    "anthropic.claude-v2:0:100k",
-    "anthropic.claude-v2:0:18k",
-    "anthropic.claude-v2:1:18k",
-    "anthropic.claude-v2:1:200k",
-    "cohere.command-r-plus-v1:0",
-    "cohere.command-r-v1:0",
-    "cohere.rerank-v3-5:0",
-    "deepseek.r1-v1:0",
-    "meta.llama3-1-70b-instruct-v1:0",
-    "meta.llama3-1-8b-instruct-v1:0",
-    "meta.llama3-2-11b-instruct-v1:0",
-    "meta.llama3-2-1b-instruct-v1:0",
-    "meta.llama3-2-3b-instruct-v1:0",
-    "meta.llama3-2-90b-instruct-v1:0",
-    "meta.llama3-3-70b-instruct-v1:0",
-    "meta.llama3-70b-instruct-v1:0",
-    "meta.llama3-8b-instruct-v1:0",
-    "meta.llama4-maverick-17b-instruct-v1:0",
-    "meta.llama4-scout-17b-instruct-v1:0",
-    "mistral.mistral-7b-instruct-v0:2",
-    "mistral.mistral-large-2402-v1:0",
-    "mistral.mistral-small-2402-v1:0",
-    "mistral.mixtral-8x7b-instruct-v0:1",
-    "mistral.pixtral-large-2502-v1:0",
-    "openai.gpt-oss-120b-1:0",
-    "openai.gpt-oss-20b-1:0",
-    "qwen.qwen3-32b-v1:0",
-    "qwen.qwen3-coder-30b-a3b-v1:0",
-    "twelvelabs.pegasus-1-2-v1:0",
-]
+from crewai.llm.constants import *  # noqa: E402, F403

lib/crewai/src/crewai/llms/hooks/__init__.py

@@ -1,6 +1,15 @@
-"""Interceptor contracts for crewai"""
+"""Deprecated: Use crewai.llm.hooks instead.
 
-from crewai.llms.hooks.base import BaseInterceptor
+.. deprecated:: 1.4.0
+"""
+
+import warnings
 
 
-__all__ = ["BaseInterceptor"]
+warnings.warn(
+    "crewai.llms.hooks is deprecated. Use crewai.llm.hooks instead.",
+    DeprecationWarning,
+    stacklevel=2,
+)
+
+from crewai.llm.hooks import *  # noqa: E402, F403

lib/crewai/src/crewai/llms/hooks/base.py

@@ -1,133 +1,15 @@
-"""Base classes for LLM transport interceptors.
+"""Deprecated: Use crewai.llm.hooks.base instead.
 
-This module provides abstract base classes for intercepting and modifying
-outbound and inbound messages at the transport level.
+.. deprecated:: 1.4.0
 """
 
-from __future__ import annotations
-
-from abc import ABC, abstractmethod
-from typing import TYPE_CHECKING, Any, Generic, TypeVar
-
-from pydantic_core import core_schema
+import warnings
 
 
-if TYPE_CHECKING:
-    from pydantic import GetCoreSchemaHandler
-    from pydantic_core import CoreSchema
+warnings.warn(
+    "crewai.llms.hooks.base is deprecated. Use crewai.llm.hooks.base instead.",
+    DeprecationWarning,
+    stacklevel=2,
+)
 
-
-T = TypeVar("T")
-U = TypeVar("U")
-
-
-class BaseInterceptor(ABC, Generic[T, U]):
-    """Abstract base class for intercepting transport-level messages.
-
-    Provides hooks to intercept and modify outbound and inbound messages
-    at the transport layer.
-
-    Type parameters:
-        T: Outbound message type (e.g., httpx.Request)
-        U: Inbound message type (e.g., httpx.Response)
-
-    Example:
-        >>> import httpx
-        >>> class CustomInterceptor(BaseInterceptor[httpx.Request, httpx.Response]):
-        ...     def on_outbound(self, message: httpx.Request) -> httpx.Request:
-        ...         message.headers["X-Custom-Header"] = "value"
-        ...         return message
-        ...
-        ...     def on_inbound(self, message: httpx.Response) -> httpx.Response:
-        ...         print(f"Status: {message.status_code}")
-        ...         return message
-    """
-
-    @abstractmethod
-    def on_outbound(self, message: T) -> T:
-        """Intercept outbound message before sending.
-
-        Args:
-            message: Outbound message object.
-
-        Returns:
-            Modified message object.
-        """
-        ...
-
-    @abstractmethod
-    def on_inbound(self, message: U) -> U:
-        """Intercept inbound message after receiving.
-
-        Args:
-            message: Inbound message object.
-
-        Returns:
-            Modified message object.
-        """
-        ...
-
-    async def aon_outbound(self, message: T) -> T:
-        """Async version of on_outbound.
-
-        Args:
-            message: Outbound message object.
-
-        Returns:
-            Modified message object.
-        """
-        raise NotImplementedError
-
-    async def aon_inbound(self, message: U) -> U:
-        """Async version of on_inbound.
-
-        Args:
-            message: Inbound message object.
-
-        Returns:
-            Modified message object.
-        """
-        raise NotImplementedError
-
-    @classmethod
-    def __get_pydantic_core_schema__(
-        cls, _source_type: Any, _handler: GetCoreSchemaHandler
-    ) -> CoreSchema:
-        """Generate Pydantic core schema for BaseInterceptor.
-
-        This allows the generic BaseInterceptor to be used in Pydantic models
-        without requiring arbitrary_types_allowed=True. The schema validates
-        that the value is an instance of BaseInterceptor.
-
-        Args:
-            _source_type: The source type being validated (unused).
-            _handler: Handler for generating schemas (unused).
-
-        Returns:
-            A Pydantic core schema that validates BaseInterceptor instances.
-        """
-        return core_schema.no_info_plain_validator_function(
-            _validate_interceptor,
-            serialization=core_schema.plain_serializer_function_ser_schema(
-                lambda x: x, return_schema=core_schema.any_schema()
-            ),
-        )
-
-
-def _validate_interceptor(value: Any) -> BaseInterceptor[T, U]:
-    """Validate that the value is a BaseInterceptor instance.
-
-    Args:
-        value: The value to validate.
-
-    Returns:
-        The validated BaseInterceptor instance.
-
-    Raises:
-        ValueError: If the value is not a BaseInterceptor instance.
-    """
-    if not isinstance(value, BaseInterceptor):
-        raise ValueError(
-            f"Expected BaseInterceptor instance, got {type(value).__name__}"
-        )
-    return value
+from crewai.llm.hooks.base import *  # noqa: E402, F403

lib/crewai/src/crewai/llms/hooks/transport.py

@@ -1,123 +1,15 @@
-"""HTTP transport implementations for LLM request/response interception.
+"""Deprecated: Use crewai.llm.hooks.transport instead.
 
-This module provides internal transport classes that integrate with BaseInterceptor
-to enable request/response modification at the transport level.
+.. deprecated:: 1.4.0
 """
 
-from __future__ import annotations
+import warnings
 
-from collections.abc import Iterable
-from typing import TYPE_CHECKING, TypedDict
 
-from httpx import (
-    AsyncHTTPTransport as _AsyncHTTPTransport,
-    HTTPTransport as _HTTPTransport,
+warnings.warn(
+    "crewai.llms.hooks.transport is deprecated. Use crewai.llm.hooks.transport instead.",
+    DeprecationWarning,
+    stacklevel=2,
 )
-from typing_extensions import NotRequired, Unpack
 
-
-if TYPE_CHECKING:
-    from ssl import SSLContext
-
-    from httpx import Limits, Request, Response
-    from httpx._types import CertTypes, ProxyTypes
-
-    from crewai.llms.hooks.base import BaseInterceptor
-
-
-class HTTPTransportKwargs(TypedDict, total=False):
-    """Typed dictionary for httpx.HTTPTransport initialization parameters.
-
-    These parameters configure the underlying HTTP transport behavior including
-    SSL verification, proxies, connection limits, and low-level socket options.
-    """
-
-    verify: bool | str | SSLContext
-    cert: NotRequired[CertTypes]
-    trust_env: bool
-    http1: bool
-    http2: bool
-    limits: Limits
-    proxy: NotRequired[ProxyTypes]
-    uds: NotRequired[str]
-    local_address: NotRequired[str]
-    retries: int
-    socket_options: NotRequired[
-        Iterable[
-            tuple[int, int, int]
-            | tuple[int, int, bytes | bytearray]
-            | tuple[int, int, None, int]
-        ]
-    ]
-
-
-class HTTPTransport(_HTTPTransport):
-    """HTTP transport that uses an interceptor for request/response modification.
-
-    This transport is used internally when a user provides a BaseInterceptor.
-    Users should not instantiate this class directly - instead, pass an interceptor
-    to the LLM client and this transport will be created automatically.
-    """
-
-    def __init__(
-        self,
-        interceptor: BaseInterceptor[Request, Response],
-        **kwargs: Unpack[HTTPTransportKwargs],
-    ) -> None:
-        """Initialize transport with interceptor.
-
-        Args:
-            interceptor: HTTP interceptor for modifying raw request/response objects.
-            **kwargs: HTTPTransport configuration parameters (verify, cert, proxy, etc.).
-        """
-        super().__init__(**kwargs)
-        self.interceptor = interceptor
-
-    def handle_request(self, request: Request) -> Response:
-        """Handle request with interception.
-
-        Args:
-            request: The HTTP request to handle.
-
-        Returns:
-            The HTTP response.
-        """
-        request = self.interceptor.on_outbound(request)
-        response = super().handle_request(request)
-        return self.interceptor.on_inbound(response)
-
-
-class AsyncHTTPTransport(_AsyncHTTPTransport):
-    """Async HTTP transport that uses an interceptor for request/response modification.
-
-    This transport is used internally when a user provides a BaseInterceptor.
-    Users should not instantiate this class directly - instead, pass an interceptor
-    to the LLM client and this transport will be created automatically.
-    """
-
-    def __init__(
-        self,
-        interceptor: BaseInterceptor[Request, Response],
-        **kwargs: Unpack[HTTPTransportKwargs],
-    ) -> None:
-        """Initialize async transport with interceptor.
-
-        Args:
-            interceptor: HTTP interceptor for modifying raw request/response objects.
-            **kwargs: HTTPTransport configuration parameters (verify, cert, proxy, etc.).
-        """
-        super().__init__(**kwargs)
-        self.interceptor = interceptor
-
-    async def handle_async_request(self, request: Request) -> Response:
-        """Handle async request with interception.
-
-        Args:
-            request: The HTTP request to handle.
-
-        Returns:
-            The HTTP response.
-        """
-        request = await self.interceptor.aon_outbound(request)
-        response = await super().handle_async_request(request)
-        return await self.interceptor.aon_inbound(response)
+from crewai.llm.hooks.transport import *  # noqa: E402, F403

lib/crewai/src/crewai/llms/internal/__init__.py

@@ -0,0 +1,15 @@
+"""Deprecated: Use crewai.llm.internal instead.
+
+.. deprecated:: 1.4.0
+"""
+
+import warnings
+
+
+warnings.warn(
+    "crewai.llms.internal is deprecated. Use crewai.llm.internal instead.",
+    DeprecationWarning,
+    stacklevel=2,
+)
+
+from crewai.llm.internal import *  # noqa: E402, F403

lib/crewai/src/crewai/llms/internal/constants.py

@@ -0,0 +1,15 @@
+"""Deprecated: Use crewai.llm.internal.constants instead.
+
+.. deprecated:: 1.4.0
+"""
+
+import warnings
+
+
+warnings.warn(
+    "crewai.llms.internal.constants is deprecated. Use crewai.llm.internal.constants instead.",
+    DeprecationWarning,
+    stacklevel=2,
+)
+
+from crewai.llm.internal.constants import *  # noqa: E402, F403

lib/crewai/src/crewai/llms/providers/__init__.py

@@ -0,0 +1,15 @@
+"""Deprecated: Use crewai.llm.providers instead.
+
+.. deprecated:: 1.4.0
+"""
+
+import warnings
+
+
+warnings.warn(
+    "crewai.llms.providers is deprecated. Use crewai.llm.providers instead.",
+    DeprecationWarning,
+    stacklevel=2,
+)
+
+from crewai.llm.providers import *  # noqa: E402, F403

lib/crewai/src/crewai/llms/third_party/__init__.py

@@ -1 +0,0 @@
-"""Third-party LLM implementations for crewAI."""

lib/crewai/src/crewai/tasks/hallucination_guardrail.py

@@ -8,7 +8,7 @@ Classes:
 
 from typing import Any
 
-from crewai.llm import LLM
+from crewai.llm.core import LLM
 from crewai.tasks.task_output import TaskOutput
 from crewai.utilities.logger import Logger
 

lib/crewai/src/crewai/tasks/llm_guardrail.py

@@ -4,7 +4,7 @@ from pydantic import BaseModel, Field
 
 from crewai.agent import Agent
 from crewai.lite_agent_output import LiteAgentOutput
-from crewai.llms.base_llm import BaseLLM
+from crewai.llm.base_llm import BaseLLM
 from crewai.tasks.task_output import TaskOutput
 
 

lib/crewai/src/crewai/tools/tool_usage.py

@@ -36,7 +36,7 @@ if TYPE_CHECKING:
     from crewai.agents.agent_builder.base_agent import BaseAgent
     from crewai.agents.tools_handler import ToolsHandler
     from crewai.lite_agent import LiteAgent
-    from crewai.llm import LLM
+    from crewai.llm.core import LLM
     from crewai.task import Task
 
 

lib/crewai/src/crewai/utilities/agent_utils.py

@@ -16,7 +16,7 @@ from crewai.agents.parser import (
     parse,
 )
 from crewai.cli.config import Settings
-from crewai.llms.base_llm import BaseLLM
+from crewai.llm.base_llm import BaseLLM
 from crewai.tools import BaseTool as CrewAITool
 from crewai.tools.base_tool import BaseTool
 from crewai.tools.structured_tool import CrewStructuredTool

lib/crewai/src/crewai/utilities/converter.py

@@ -19,7 +19,7 @@ if TYPE_CHECKING:
     from crewai.agent import Agent
     from crewai.agents.agent_builder.base_agent import BaseAgent
     from crewai.llm import LLM
-    from crewai.llms.base_llm import BaseLLM
+    from crewai.llm.base_llm import BaseLLM
 
 _JSON_PATTERN: Final[re.Pattern[str]] = re.compile(r"({.*})", re.DOTALL)
 _I18N = get_i18n()

lib/crewai/src/crewai/utilities/evaluators/crew_evaluator_handler.py

@@ -11,7 +11,7 @@ from rich.table import Table
 from crewai.agent import Agent
 from crewai.events.event_bus import crewai_event_bus
 from crewai.events.types.crew_events import CrewTestResultEvent
-from crewai.llms.base_llm import BaseLLM
+from crewai.llm.base_llm import BaseLLM
 from crewai.task import Task
 from crewai.tasks.task_output import TaskOutput
 

lib/crewai/src/crewai/utilities/internal_instructor.py

@@ -10,7 +10,7 @@ from crewai.utilities.logger_utils import suppress_warnings
 if TYPE_CHECKING:
     from crewai.agent import Agent
     from crewai.llm import LLM
-    from crewai.llms.base_llm import BaseLLM
+    from crewai.llm.base_llm import BaseLLM
     from crewai.utilities.types import LLMMessage
 
 

lib/crewai/src/crewai/utilities/llm_utils.py

@@ -4,7 +4,7 @@ from typing import Any, Final
 
 from crewai.cli.constants import DEFAULT_LLM_MODEL, ENV_VARS, LITELLM_PARAMS
 from crewai.llm import LLM
-from crewai.llms.base_llm import BaseLLM
+from crewai.llm.base_llm import BaseLLM
 
 
 logger = logging.getLogger(__name__)

lib/crewai/src/crewai/utilities/planning_handler.py

@@ -5,7 +5,7 @@ import logging
 from pydantic import BaseModel, Field
 
 from crewai.agent import Agent
-from crewai.llms.base_llm import BaseLLM
+from crewai.llm.base_llm import BaseLLM
 from crewai.task import Task
 
 

lib/crewai/src/crewai/utilities/tool_utils.py

@@ -22,7 +22,7 @@ if TYPE_CHECKING:
     from crewai.agents.agent_builder.base_agent import BaseAgent
     from crewai.crew import Crew
     from crewai.llm import LLM
-    from crewai.llms.base_llm import BaseLLM
+    from crewai.llm.base_llm import BaseLLM
     from crewai.task import Task
 
 

lib/crewai/tests/agents/test_agent.py

@@ -14,7 +14,7 @@ from crewai.knowledge.knowledge_config import KnowledgeConfig
 from crewai.knowledge.source.base_knowledge_source import BaseKnowledgeSource
 from crewai.knowledge.source.string_knowledge_source import StringKnowledgeSource
 from crewai.llm import LLM
-from crewai.llms.base_llm import BaseLLM
+from crewai.llm.base_llm import BaseLLM
 from crewai.process import Process
 from crewai.tools.tool_calling import InstructorToolCalling
 from crewai.tools.tool_usage import ToolUsage

lib/crewai/tests/agents/test_lite_agent.py

@@ -9,7 +9,7 @@ from crewai.events.types.agent_events import LiteAgentExecutionStartedEvent
 from crewai.events.types.tool_usage_events import ToolUsageStartedEvent
 from crewai.lite_agent import LiteAgent
 from crewai.lite_agent_output import LiteAgentOutput
-from crewai.llms.base_llm import BaseLLM
+from crewai.llm.base_llm import BaseLLM
 from pydantic import BaseModel, Field
 import pytest
 

lib/crewai/tests/cassettes/test_agent_function_calling_llm.yaml

@@ -590,7 +590,7 @@ interactions:
       "<function BaseTool.<lambda> at 0x107389260>", "result_as_answer": "False",
       "max_usage_count": "None", "current_usage_count": "0"}], "max_iter": 2, "agent_executor":
       "<crewai.agents.crew_agent_executor.CrewAgentExecutor object at 0x130de6540>",
-      "llm": "<crewai.llms.providers.openai.completion.OpenAICompletion object at
+      "llm": "<crewai.llm.providers.openai.completion.OpenAICompletion object at
       0x130db6de0>", "crew": {"parent_flow": null, "name": "crew", "cache": true,
       "tasks": ["{''used_tools'': 0, ''tools_errors'': 0, ''delegations'': 0, ''i18n'':
       {''prompt_file'': None}, ''name'': None, ''prompt_context'': '''', ''description'':
@@ -605,7 +605,7 @@ interactions:
       ''description_updated'': False, ''cache_function'': <function BaseTool.<lambda>
       at 0x107389260>, ''result_as_answer'': False, ''max_usage_count'': None, ''current_usage_count'':
       0}], ''max_iter'': 2, ''agent_executor'': <crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x130de6540>, ''llm'': <crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x130de6540>, ''llm'': <crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x130db6de0>, ''crew'': Crew(id=991ac83f-9a29-411f-b0a0-0a335c7a2d0e,
       process=Process.sequential, number_of_agents=1, number_of_tasks=1), ''i18n'':
       {''prompt_file'': None}, ''cache_handler'': {}, ''tools_handler'': <crewai.agents.tools_handler.ToolsHandler
@@ -634,7 +634,7 @@ interactions:
       ''abc.Learn_About_Ai''>, ''description_updated'': False, ''cache_function'':
       <function BaseTool.<lambda> at 0x107389260>, ''result_as_answer'': False, ''max_usage_count'':
       None, ''current_usage_count'': 0}], ''max_iter'': 2, ''agent_executor'': <crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x130de6540>, ''llm'': <crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x130de6540>, ''llm'': <crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x130db6de0>, ''crew'': Crew(id=991ac83f-9a29-411f-b0a0-0a335c7a2d0e,
       process=Process.sequential, number_of_agents=1, number_of_tasks=1), ''i18n'':
       {''prompt_file'': None}, ''cache_handler'': {}, ''tools_handler'': <crewai.agents.tools_handler.ToolsHandler
@@ -657,7 +657,7 @@ interactions:
       {"fingerprint": {"metadata": "{}"}}, "callbacks": [], "adapted_agent": false,
       "knowledge_config": null, "max_execution_time": null, "agent_ops_agent_name":
       "test role", "agent_ops_agent_id": null, "step_callback": null, "use_system_prompt":
-      true, "function_calling_llm": "<crewai.llms.providers.openai.completion.OpenAICompletion
+      true, "function_calling_llm": "<crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x130db7020>", "system_template": null, "prompt_template": null, "response_template":
       null, "allow_code_execution": false, "respect_context_window": true, "max_retry_limit":
       2, "multimodal": false, "inject_date": false, "date_format": "%Y-%m-%d", "code_execution_mode":
@@ -1068,7 +1068,7 @@ interactions:
       "<function BaseTool.<lambda> at 0x107e394e0>", "result_as_answer": "False",
       "max_usage_count": "None", "current_usage_count": "0"}], "max_iter": 2, "agent_executor":
       "<crewai.agents.crew_agent_executor.CrewAgentExecutor object at 0x13b37c980>",
-      "llm": "<crewai.llms.providers.openai.completion.OpenAICompletion object at
+      "llm": "<crewai.llm.providers.openai.completion.OpenAICompletion object at
       0x13b7563c0>", "crew": {"parent_flow": null, "name": "crew", "cache": true,
       "tasks": ["{''used_tools'': 0, ''tools_errors'': 0, ''delegations'': 0, ''i18n'':
       {''prompt_file'': None}, ''name'': None, ''prompt_context'': '''', ''description'':
@@ -1083,7 +1083,7 @@ interactions:
       ''description_updated'': False, ''cache_function'': <function BaseTool.<lambda>
       at 0x107e394e0>, ''result_as_answer'': False, ''max_usage_count'': None, ''current_usage_count'':
       0}], ''max_iter'': 2, ''agent_executor'': <crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x13b37c980>, ''llm'': <crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x13b37c980>, ''llm'': <crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x13b7563c0>, ''crew'': Crew(id=f38365e9-3206-45b6-8754-950cb03fe57e,
       process=Process.sequential, number_of_agents=1, number_of_tasks=1), ''i18n'':
       {''prompt_file'': None}, ''cache_handler'': {}, ''tools_handler'': <crewai.agents.tools_handler.ToolsHandler
@@ -1112,7 +1112,7 @@ interactions:
       ''abc.Learn_About_Ai''>, ''description_updated'': False, ''cache_function'':
       <function BaseTool.<lambda> at 0x107e394e0>, ''result_as_answer'': False, ''max_usage_count'':
       None, ''current_usage_count'': 0}], ''max_iter'': 2, ''agent_executor'': <crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x13b37c980>, ''llm'': <crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x13b37c980>, ''llm'': <crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x13b7563c0>, ''crew'': Crew(id=f38365e9-3206-45b6-8754-950cb03fe57e,
       process=Process.sequential, number_of_agents=1, number_of_tasks=1), ''i18n'':
       {''prompt_file'': None}, ''cache_handler'': {}, ''tools_handler'': <crewai.agents.tools_handler.ToolsHandler
@@ -1135,7 +1135,7 @@ interactions:
       {"fingerprint": {"metadata": "{}"}}, "callbacks": [], "adapted_agent": false,
       "knowledge_config": null, "max_execution_time": null, "agent_ops_agent_name":
       "test role", "agent_ops_agent_id": null, "step_callback": null, "use_system_prompt":
-      true, "function_calling_llm": "<crewai.llms.providers.openai.completion.OpenAICompletion
+      true, "function_calling_llm": "<crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x13b756690>", "system_template": null, "prompt_template": null, "response_template":
       null, "allow_code_execution": false, "respect_context_window": true, "max_retry_limit":
       2, "multimodal": false, "inject_date": false, "date_format": "%Y-%m-%d", "code_execution_mode":

lib/crewai/tests/cassettes/test_agent_remembers_output_format_after_using_tools_too_many_times.yaml

@@ -1274,7 +1274,7 @@ interactions:
       "b6cf723e-04c8-40c5-a927-e2078cfbae59", "role": "test role", "goal": "test goal",
       "backstory": "test backstory", "cache": true, "verbose": true, "max_rpm": null,
       "allow_delegation": false, "tools": [], "max_iter": 6, "agent_executor": "<crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x13ab0abd0>", "llm": "<crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x13ab0abd0>", "llm": "<crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x13ab0b050>", "crew": {"parent_flow": null, "name": "crew", "cache":
       true, "tasks": ["{''used_tools'': 0, ''tools_errors'': 0, ''delegations'': 0,
       ''i18n'': {''prompt_file'': None}, ''name'': None, ''prompt_context'': '''',
@@ -1285,7 +1285,7 @@ interactions:
       ''test role'', ''goal'': ''test goal'', ''backstory'': ''test backstory'', ''cache'':
       True, ''verbose'': True, ''max_rpm'': None, ''allow_delegation'': False, ''tools'':
       [], ''max_iter'': 6, ''agent_executor'': <crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x13ab0abd0>, ''llm'': <crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x13ab0abd0>, ''llm'': <crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x13ab0b050>, ''crew'': Crew(id=004dd8a0-dd87-43fa-bdc8-07f449808028,
       process=Process.sequential, number_of_agents=1, number_of_tasks=1), ''i18n'':
       {''prompt_file'': None}, ''cache_handler'': {}, ''tools_handler'': <crewai.agents.tools_handler.ToolsHandler
@@ -1309,7 +1309,7 @@ interactions:
       role'', ''goal'': ''test goal'', ''backstory'': ''test backstory'', ''cache'':
       True, ''verbose'': True, ''max_rpm'': None, ''allow_delegation'': False, ''tools'':
       [], ''max_iter'': 6, ''agent_executor'': <crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x13ab0abd0>, ''llm'': <crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x13ab0abd0>, ''llm'': <crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x13ab0b050>, ''crew'': Crew(id=004dd8a0-dd87-43fa-bdc8-07f449808028,
       process=Process.sequential, number_of_agents=1, number_of_tasks=1), ''i18n'':
       {''prompt_file'': None}, ''cache_handler'': {}, ''tools_handler'': <crewai.agents.tools_handler.ToolsHandler
@@ -1491,7 +1491,7 @@ interactions:
       "goal": "test goal", "backstory": "test backstory", "cache": true, "verbose":
       true, "max_rpm": null, "allow_delegation": false, "tools": [], "max_iter": 6,
       "agent_executor": "<crewai.agents.crew_agent_executor.CrewAgentExecutor object
-      at 0x13ab0abd0>", "llm": "<crewai.llms.providers.openai.completion.OpenAICompletion
+      at 0x13ab0abd0>", "llm": "<crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x13ab0b050>", "crew": {"parent_flow": null, "name": "crew", "cache":
       true, "tasks": ["{''used_tools'': 2, ''tools_errors'': 0, ''delegations'': 0,
       ''i18n'': {''prompt_file'': None}, ''name'': None, ''prompt_context'': '''',
@@ -1502,7 +1502,7 @@ interactions:
       ''test role'', ''goal'': ''test goal'', ''backstory'': ''test backstory'', ''cache'':
       True, ''verbose'': True, ''max_rpm'': None, ''allow_delegation'': False, ''tools'':
       [], ''max_iter'': 6, ''agent_executor'': <crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x13ab0abd0>, ''llm'': <crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x13ab0abd0>, ''llm'': <crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x13ab0b050>, ''crew'': Crew(id=004dd8a0-dd87-43fa-bdc8-07f449808028,
       process=Process.sequential, number_of_agents=1, number_of_tasks=1), ''i18n'':
       {''prompt_file'': None}, ''cache_handler'': {}, ''tools_handler'': <crewai.agents.tools_handler.ToolsHandler
@@ -1528,7 +1528,7 @@ interactions:
       role'', ''goal'': ''test goal'', ''backstory'': ''test backstory'', ''cache'':
       True, ''verbose'': True, ''max_rpm'': None, ''allow_delegation'': False, ''tools'':
       [], ''max_iter'': 6, ''agent_executor'': <crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x13ab0abd0>, ''llm'': <crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x13ab0abd0>, ''llm'': <crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x13ab0b050>, ''crew'': Crew(id=004dd8a0-dd87-43fa-bdc8-07f449808028,
       process=Process.sequential, number_of_agents=1, number_of_tasks=1), ''i18n'':
       {''prompt_file'': None}, ''cache_handler'': {}, ''tools_handler'': <crewai.agents.tools_handler.ToolsHandler
@@ -1660,7 +1660,7 @@ interactions:
       role", "goal": "test goal", "backstory": "test backstory", "cache": true, "verbose":
       true, "max_rpm": null, "allow_delegation": false, "tools": [], "max_iter": 6,
       "agent_executor": "<crewai.agents.crew_agent_executor.CrewAgentExecutor object
-      at 0x13ab0abd0>", "llm": "<crewai.llms.providers.openai.completion.OpenAICompletion
+      at 0x13ab0abd0>", "llm": "<crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x13ab0b050>", "crew": {"parent_flow": null, "name": "crew", "cache":
       true, "tasks": ["{''used_tools'': 3, ''tools_errors'': 0, ''delegations'': 0,
       ''i18n'': {''prompt_file'': None}, ''name'': None, ''prompt_context'': '''',
@@ -1671,7 +1671,7 @@ interactions:
       ''test role'', ''goal'': ''test goal'', ''backstory'': ''test backstory'', ''cache'':
       True, ''verbose'': True, ''max_rpm'': None, ''allow_delegation'': False, ''tools'':
       [], ''max_iter'': 6, ''agent_executor'': <crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x13ab0abd0>, ''llm'': <crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x13ab0abd0>, ''llm'': <crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x13ab0b050>, ''crew'': Crew(id=004dd8a0-dd87-43fa-bdc8-07f449808028,
       process=Process.sequential, number_of_agents=1, number_of_tasks=1), ''i18n'':
       {''prompt_file'': None}, ''cache_handler'': {}, ''tools_handler'': <crewai.agents.tools_handler.ToolsHandler
@@ -1698,7 +1698,7 @@ interactions:
       role'', ''goal'': ''test goal'', ''backstory'': ''test backstory'', ''cache'':
       True, ''verbose'': True, ''max_rpm'': None, ''allow_delegation'': False, ''tools'':
       [], ''max_iter'': 6, ''agent_executor'': <crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x13ab0abd0>, ''llm'': <crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x13ab0abd0>, ''llm'': <crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x13ab0b050>, ''crew'': Crew(id=004dd8a0-dd87-43fa-bdc8-07f449808028,
       process=Process.sequential, number_of_agents=1, number_of_tasks=1), ''i18n'':
       {''prompt_file'': None}, ''cache_handler'': {}, ''tools_handler'': <crewai.agents.tools_handler.ToolsHandler
@@ -1839,7 +1839,7 @@ interactions:
       "goal": "test goal", "backstory": "test backstory", "cache": true, "verbose":
       true, "max_rpm": null, "allow_delegation": false, "tools": [], "max_iter": 6,
       "agent_executor": "<crewai.agents.crew_agent_executor.CrewAgentExecutor object
-      at 0x13ab0abd0>", "llm": "<crewai.llms.providers.openai.completion.OpenAICompletion
+      at 0x13ab0abd0>", "llm": "<crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x13ab0b050>", "crew": {"parent_flow": null, "name": "crew", "cache":
       true, "tasks": ["{''used_tools'': 4, ''tools_errors'': 0, ''delegations'': 0,
       ''i18n'': {''prompt_file'': None}, ''name'': None, ''prompt_context'': '''',
@@ -1850,7 +1850,7 @@ interactions:
       ''test role'', ''goal'': ''test goal'', ''backstory'': ''test backstory'', ''cache'':
       True, ''verbose'': True, ''max_rpm'': None, ''allow_delegation'': False, ''tools'':
       [], ''max_iter'': 6, ''agent_executor'': <crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x13ab0abd0>, ''llm'': <crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x13ab0abd0>, ''llm'': <crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x13ab0b050>, ''crew'': Crew(id=004dd8a0-dd87-43fa-bdc8-07f449808028,
       process=Process.sequential, number_of_agents=1, number_of_tasks=1), ''i18n'':
       {''prompt_file'': None}, ''cache_handler'': {}, ''tools_handler'': <crewai.agents.tools_handler.ToolsHandler
@@ -1879,7 +1879,7 @@ interactions:
       role'', ''goal'': ''test goal'', ''backstory'': ''test backstory'', ''cache'':
       True, ''verbose'': True, ''max_rpm'': None, ''allow_delegation'': False, ''tools'':
       [], ''max_iter'': 6, ''agent_executor'': <crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x13ab0abd0>, ''llm'': <crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x13ab0abd0>, ''llm'': <crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x13ab0b050>, ''crew'': Crew(id=004dd8a0-dd87-43fa-bdc8-07f449808028,
       process=Process.sequential, number_of_agents=1, number_of_tasks=1), ''i18n'':
       {''prompt_file'': None}, ''cache_handler'': {}, ''tools_handler'': <crewai.agents.tools_handler.ToolsHandler
@@ -2029,7 +2029,7 @@ interactions:
       "goal": "test goal", "backstory": "test backstory", "cache": true, "verbose":
       true, "max_rpm": null, "allow_delegation": false, "tools": [], "max_iter": 6,
       "agent_executor": "<crewai.agents.crew_agent_executor.CrewAgentExecutor object
-      at 0x13ab0abd0>", "llm": "<crewai.llms.providers.openai.completion.OpenAICompletion
+      at 0x13ab0abd0>", "llm": "<crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x13ab0b050>", "crew": {"parent_flow": null, "name": "crew", "cache":
       true, "tasks": ["{''used_tools'': 5, ''tools_errors'': 0, ''delegations'': 0,
       ''i18n'': {''prompt_file'': None}, ''name'': None, ''prompt_context'': '''',
@@ -2040,7 +2040,7 @@ interactions:
       ''test role'', ''goal'': ''test goal'', ''backstory'': ''test backstory'', ''cache'':
       True, ''verbose'': True, ''max_rpm'': None, ''allow_delegation'': False, ''tools'':
       [], ''max_iter'': 6, ''agent_executor'': <crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x13ab0abd0>, ''llm'': <crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x13ab0abd0>, ''llm'': <crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x13ab0b050>, ''crew'': Crew(id=004dd8a0-dd87-43fa-bdc8-07f449808028,
       process=Process.sequential, number_of_agents=1, number_of_tasks=1), ''i18n'':
       {''prompt_file'': None}, ''cache_handler'': {}, ''tools_handler'': <crewai.agents.tools_handler.ToolsHandler
@@ -2070,7 +2070,7 @@ interactions:
       role'', ''goal'': ''test goal'', ''backstory'': ''test backstory'', ''cache'':
       True, ''verbose'': True, ''max_rpm'': None, ''allow_delegation'': False, ''tools'':
       [], ''max_iter'': 6, ''agent_executor'': <crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x13ab0abd0>, ''llm'': <crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x13ab0abd0>, ''llm'': <crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x13ab0b050>, ''crew'': Crew(id=004dd8a0-dd87-43fa-bdc8-07f449808028,
       process=Process.sequential, number_of_agents=1, number_of_tasks=1), ''i18n'':
       {''prompt_file'': None}, ''cache_handler'': {}, ''tools_handler'': <crewai.agents.tools_handler.ToolsHandler

lib/crewai/tests/cassettes/test_agent_respect_the_max_rpm_set_over_crew_rpm.yaml

@@ -1082,7 +1082,7 @@ interactions:
       "role": "test role", "goal": "test goal", "backstory": "test backstory", "cache":
       true, "verbose": true, "max_rpm": 10, "allow_delegation": false, "tools": [],
       "max_iter": 4, "agent_executor": "<crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x133d41100>", "llm": "<crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x133d41100>", "llm": "<crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x133d40500>", "crew": {"parent_flow": null, "name": "crew", "cache":
       true, "tasks": ["{''used_tools'': 0, ''tools_errors'': 0, ''delegations'': 0,
       ''i18n'': {''prompt_file'': None}, ''name'': None, ''prompt_context'': '''',
@@ -1093,7 +1093,7 @@ interactions:
       ''test role'', ''goal'': ''test goal'', ''backstory'': ''test backstory'', ''cache'':
       True, ''verbose'': True, ''max_rpm'': 10, ''allow_delegation'': False, ''tools'':
       [], ''max_iter'': 4, ''agent_executor'': <crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x133d41100>, ''llm'': <crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x133d41100>, ''llm'': <crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x133d40500>, ''crew'': Crew(id=4c6d502e-f6ec-446a-8f76-644563c4aa94,
       process=Process.sequential, number_of_agents=1, number_of_tasks=1), ''i18n'':
       {''prompt_file'': None}, ''cache_handler'': {}, ''tools_handler'': <crewai.agents.tools_handler.ToolsHandler
@@ -1117,7 +1117,7 @@ interactions:
       ''test role'', ''goal'': ''test goal'', ''backstory'': ''test backstory'', ''cache'':
       True, ''verbose'': True, ''max_rpm'': 10, ''allow_delegation'': False, ''tools'':
       [], ''max_iter'': 4, ''agent_executor'': <crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x133d41100>, ''llm'': <crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x133d41100>, ''llm'': <crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x133d40500>, ''crew'': Crew(id=4c6d502e-f6ec-446a-8f76-644563c4aa94,
       process=Process.sequential, number_of_agents=1, number_of_tasks=1), ''i18n'':
       {''prompt_file'': None}, ''cache_handler'': {}, ''tools_handler'': <crewai.agents.tools_handler.ToolsHandler
@@ -1910,7 +1910,7 @@ interactions:
       "role": "test role", "goal": "test goal", "backstory": "test backstory", "cache":
       true, "verbose": true, "max_rpm": 10, "allow_delegation": false, "tools": [],
       "max_iter": 4, "agent_executor": "<crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x10308d610>", "llm": "<crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x10308d610>", "llm": "<crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x129201640>", "crew": {"parent_flow": null, "name": "crew", "cache":
       true, "tasks": ["{''used_tools'': 0, ''tools_errors'': 0, ''delegations'': 0,
       ''i18n'': {''prompt_file'': None}, ''name'': None, ''prompt_context'': '''',
@@ -1921,7 +1921,7 @@ interactions:
       ''test role'', ''goal'': ''test goal'', ''backstory'': ''test backstory'', ''cache'':
       True, ''verbose'': True, ''max_rpm'': 10, ''allow_delegation'': False, ''tools'':
       [], ''max_iter'': 4, ''agent_executor'': <crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x10308d610>, ''llm'': <crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x10308d610>, ''llm'': <crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x129201640>, ''crew'': Crew(id=1a07d718-fed5-49fa-bee2-de2db91c9f33,
       process=Process.sequential, number_of_agents=1, number_of_tasks=1), ''i18n'':
       {''prompt_file'': None}, ''cache_handler'': {}, ''tools_handler'': <crewai.agents.tools_handler.ToolsHandler
@@ -1945,7 +1945,7 @@ interactions:
       ''test role'', ''goal'': ''test goal'', ''backstory'': ''test backstory'', ''cache'':
       True, ''verbose'': True, ''max_rpm'': 10, ''allow_delegation'': False, ''tools'':
       [], ''max_iter'': 4, ''agent_executor'': <crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x10308d610>, ''llm'': <crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x10308d610>, ''llm'': <crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x129201640>, ''crew'': Crew(id=1a07d718-fed5-49fa-bee2-de2db91c9f33,
       process=Process.sequential, number_of_agents=1, number_of_tasks=1), ''i18n'':
       {''prompt_file'': None}, ''cache_handler'': {}, ''tools_handler'': <crewai.agents.tools_handler.ToolsHandler

lib/crewai/tests/cassettes/test_do_not_allow_crewai_trigger_context_for_first_task_hierarchical.yaml

@@ -937,7 +937,7 @@ interactions:
       "description_updated": "False", "cache_function": "<function BaseTool.<lambda>
       at 0x10614d3a0>", "result_as_answer": "False", "max_usage_count": "None", "current_usage_count":
       "0"}], "max_iter": 25, "agent_executor": "<crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x10f6c3bc0>", "llm": "<crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x10f6c3bc0>", "llm": "<crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x10f6c27e0>", "crew": {"parent_flow": null, "name": "crew", "cache":
       true, "tasks": ["{''used_tools'': 0, ''tools_errors'': 0, ''delegations'': 0,
       ''i18n'': {''prompt_file'': None}, ''name'': None, ''prompt_context'': '''',
@@ -977,7 +977,7 @@ interactions:
       ''description_updated'': False, ''cache_function'': <function BaseTool.<lambda>
       at 0x10614d3a0>, ''result_as_answer'': False, ''max_usage_count'': None, ''current_usage_count'':
       0}], ''max_iter'': 25, ''agent_executor'': <crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x10f6c3bc0>, ''llm'': <crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x10f6c3bc0>, ''llm'': <crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x10f6c27e0>, ''crew'': Crew(id=49cbb747-f055-4636-bbca-9e8a450c05f6,
       process=Process.hierarchical, number_of_agents=2, number_of_tasks=1), ''i18n'':
       {''prompt_file'': None}, ''cache_handler'': {}, ''tools_handler'': <crewai.agents.tools_handler.ToolsHandler
@@ -996,7 +996,7 @@ interactions:
       ''First Agent'', ''goal'': ''First goal'', ''backstory'': ''First backstory'',
       ''cache'': True, ''verbose'': False, ''max_rpm'': None, ''allow_delegation'':
       False, ''tools'': [], ''max_iter'': 25, ''agent_executor'': <crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x11059ca10>, ''llm'': <crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x11059ca10>, ''llm'': <crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x10f6e6ae0>, ''crew'': Crew(id=49cbb747-f055-4636-bbca-9e8a450c05f6,
       process=Process.hierarchical, number_of_agents=2, number_of_tasks=1), ''i18n'':
       {''prompt_file'': None}, ''cache_handler'': {}, ''tools_handler'': <crewai.agents.tools_handler.ToolsHandler
@@ -1007,7 +1007,7 @@ interactions:
       ''role'': ''Second Agent'', ''goal'': ''Second goal'', ''backstory'': ''Second
       backstory'', ''cache'': True, ''verbose'': False, ''max_rpm'': None, ''allow_delegation'':
       False, ''tools'': [], ''max_iter'': 25, ''agent_executor'': <crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x10f6c3500>, ''llm'': <crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x10f6c3500>, ''llm'': <crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x10f6d2000>, ''crew'': Crew(id=49cbb747-f055-4636-bbca-9e8a450c05f6,
       process=Process.hierarchical, number_of_agents=2, number_of_tasks=1), ''i18n'':
       {''prompt_file'': None}, ''cache_handler'': {}, ''tools_handler'': <crewai.agents.tools_handler.ToolsHandler
@@ -1017,7 +1017,7 @@ interactions:
       False, ''knowledge_config'': None}"], "process": "hierarchical", "verbose":
       false, "memory": false, "short_term_memory": null, "long_term_memory": null,
       "entity_memory": null, "external_memory": null, "embedder": null, "usage_metrics":
-      null, "manager_llm": "<crewai.llms.providers.openai.completion.OpenAICompletion
+      null, "manager_llm": "<crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x10f6c27e0>", "manager_agent": {"id": "UUID(''b0898472-5e3b-45bb-bd90-05bad0b5a8ce'')",
       "role": "''Crew Manager''", "goal": "''Manage the team to complete the task
       in the best way possible.''", "backstory": "\"You are a seasoned manager with
@@ -1053,7 +1053,7 @@ interactions:
       ''description_updated'': False, ''cache_function'': <function BaseTool.<lambda>
       at 0x10614d3a0>, ''result_as_answer'': False, ''max_usage_count'': None, ''current_usage_count'':
       0}]", "max_iter": "25", "agent_executor": "<crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x10f6c3bc0>", "llm": "<crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x10f6c3bc0>", "llm": "<crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x10f6c27e0>", "crew": "Crew(id=49cbb747-f055-4636-bbca-9e8a450c05f6,
       process=Process.hierarchical, number_of_agents=2, number_of_tasks=1)", "i18n":
       "{''prompt_file'': None}", "cache_handler": "{}", "tools_handler": "<crewai.agents.tools_handler.ToolsHandler
@@ -1805,7 +1805,7 @@ interactions:
       "description_updated": "False", "cache_function": "<function BaseTool.<lambda>
       at 0x107e394e0>", "result_as_answer": "False", "max_usage_count": "None", "current_usage_count":
       "0"}], "max_iter": 25, "agent_executor": "<crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x1388bedb0>", "llm": "<crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x1388bedb0>", "llm": "<crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x1388bf710>", "crew": {"parent_flow": null, "name": "crew", "cache":
       true, "tasks": ["{''used_tools'': 0, ''tools_errors'': 0, ''delegations'': 0,
       ''i18n'': {''prompt_file'': None}, ''name'': None, ''prompt_context'': '''',
@@ -1845,7 +1845,7 @@ interactions:
       ''description_updated'': False, ''cache_function'': <function BaseTool.<lambda>
       at 0x107e394e0>, ''result_as_answer'': False, ''max_usage_count'': None, ''current_usage_count'':
       0}], ''max_iter'': 25, ''agent_executor'': <crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x1388bedb0>, ''llm'': <crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x1388bedb0>, ''llm'': <crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x1388bf710>, ''crew'': Crew(id=4d744f3e-0589-4d1d-b1c1-6aa8b52478ac,
       process=Process.hierarchical, number_of_agents=2, number_of_tasks=1), ''i18n'':
       {''prompt_file'': None}, ''cache_handler'': {}, ''tools_handler'': <crewai.agents.tools_handler.ToolsHandler
@@ -1864,7 +1864,7 @@ interactions:
       ''First Agent'', ''goal'': ''First goal'', ''backstory'': ''First backstory'',
       ''cache'': True, ''verbose'': False, ''max_rpm'': None, ''allow_delegation'':
       False, ''tools'': [], ''max_iter'': 25, ''agent_executor'': <crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x1388d5c70>, ''llm'': <crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x1388d5c70>, ''llm'': <crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x1388bde80>, ''crew'': Crew(id=4d744f3e-0589-4d1d-b1c1-6aa8b52478ac,
       process=Process.hierarchical, number_of_agents=2, number_of_tasks=1), ''i18n'':
       {''prompt_file'': None}, ''cache_handler'': {}, ''tools_handler'': <crewai.agents.tools_handler.ToolsHandler
@@ -1875,7 +1875,7 @@ interactions:
       ''role'': ''Second Agent'', ''goal'': ''Second goal'', ''backstory'': ''Second
       backstory'', ''cache'': True, ''verbose'': False, ''max_rpm'': None, ''allow_delegation'':
       False, ''tools'': [], ''max_iter'': 25, ''agent_executor'': <crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x1388bf7d0>, ''llm'': <crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x1388bf7d0>, ''llm'': <crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x1388bfb90>, ''crew'': Crew(id=4d744f3e-0589-4d1d-b1c1-6aa8b52478ac,
       process=Process.hierarchical, number_of_agents=2, number_of_tasks=1), ''i18n'':
       {''prompt_file'': None}, ''cache_handler'': {}, ''tools_handler'': <crewai.agents.tools_handler.ToolsHandler
@@ -1885,7 +1885,7 @@ interactions:
       False, ''knowledge_config'': None}"], "process": "hierarchical", "verbose":
       false, "memory": false, "short_term_memory": null, "long_term_memory": null,
       "entity_memory": null, "external_memory": null, "embedder": null, "usage_metrics":
-      null, "manager_llm": "<crewai.llms.providers.openai.completion.OpenAICompletion
+      null, "manager_llm": "<crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x1388bf710>", "manager_agent": {"id": "UUID(''09794b42-447f-4b7a-b634-3a861f457357'')",
       "role": "''Crew Manager''", "goal": "''Manage the team to complete the task
       in the best way possible.''", "backstory": "\"You are a seasoned manager with
@@ -1921,7 +1921,7 @@ interactions:
       ''description_updated'': False, ''cache_function'': <function BaseTool.<lambda>
       at 0x107e394e0>, ''result_as_answer'': False, ''max_usage_count'': None, ''current_usage_count'':
       0}]", "max_iter": "25", "agent_executor": "<crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x1388bedb0>", "llm": "<crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x1388bedb0>", "llm": "<crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x1388bf710>", "crew": "Crew(id=4d744f3e-0589-4d1d-b1c1-6aa8b52478ac,
       process=Process.hierarchical, number_of_agents=2, number_of_tasks=1)", "i18n":
       "{''prompt_file'': None}", "cache_handler": "{}", "tools_handler": "<crewai.agents.tools_handler.ToolsHandler

lib/crewai/tests/cassettes/test_sets_parent_flow_when_inside_flow.yaml

@@ -126,7 +126,7 @@ interactions:
       a freelancer and is now working on doing research and analysis for a new customer.\",
       ''cache'': True, ''verbose'': False, ''max_rpm'': None, ''allow_delegation'':
       False, ''tools'': [], ''max_iter'': 25, ''agent_executor'': <crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x12b973fe0>, ''llm'': <crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x12b973fe0>, ''llm'': <crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x12b910290>, ''crew'': None, ''i18n'': {''prompt_file'': None}, ''cache_handler'':
       {}, ''tools_handler'': <crewai.agents.tools_handler.ToolsHandler object at 0x12b9934d0>,
       ''tools_results'': [], ''max_tokens'': None, ''knowledge'': None, ''knowledge_sources'':
@@ -149,7 +149,7 @@ interactions:
       writing content for a new customer.\", ''cache'': True, ''verbose'': False,
       ''max_rpm'': None, ''allow_delegation'': False, ''tools'': [], ''max_iter'':
       25, ''agent_executor'': <crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x12b7bbbf0>, ''llm'': <crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x12b7bbbf0>, ''llm'': <crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x12b9903b0>, ''crew'': None, ''i18n'': {''prompt_file'': None}, ''cache_handler'':
       {}, ''tools_handler'': <crewai.agents.tools_handler.ToolsHandler object at 0x12b631bb0>,
       ''tools_results'': [], ''max_tokens'': None, ''knowledge'': None, ''knowledge_sources'':
@@ -169,7 +169,7 @@ interactions:
       a freelancer and is now working on doing research and analysis for a new customer.\"",
       "cache": "True", "verbose": "False", "max_rpm": "None", "allow_delegation":
       "False", "tools": "[]", "max_iter": "25", "agent_executor": "<crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x12b973fe0>", "llm": "<crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x12b973fe0>", "llm": "<crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x12b910290>", "crew": "None", "i18n": "{''prompt_file'': None}",
       "cache_handler": "{}", "tools_handler": "<crewai.agents.tools_handler.ToolsHandler
       object at 0x12b9934d0>", "tools_results": "[]", "max_tokens": "None", "knowledge":
@@ -182,7 +182,7 @@ interactions:
       You work as a freelancer and are now working on writing content for a new customer.\"",
       "cache": "True", "verbose": "False", "max_rpm": "None", "allow_delegation":
       "False", "tools": "[]", "max_iter": "25", "agent_executor": "<crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x12b7bbbf0>", "llm": "<crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x12b7bbbf0>", "llm": "<crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x12b9903b0>", "crew": "None", "i18n": "{''prompt_file'': None}",
       "cache_handler": "{}", "tools_handler": "<crewai.agents.tools_handler.ToolsHandler
       object at 0x12b631bb0>", "tools_results": "[]", "max_tokens": "None", "knowledge":
@@ -214,7 +214,7 @@ interactions:
       a freelancer and is now working on doing research and analysis for a new customer.\",
       ''cache'': True, ''verbose'': False, ''max_rpm'': None, ''allow_delegation'':
       False, ''tools'': [], ''max_iter'': 25, ''agent_executor'': <crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x12b973fe0>, ''llm'': <crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x12b973fe0>, ''llm'': <crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x12b910290>, ''crew'': None, ''i18n'': {''prompt_file'': None}, ''cache_handler'':
       {}, ''tools_handler'': <crewai.agents.tools_handler.ToolsHandler object at 0x12b9934d0>,
       ''tools_results'': [], ''max_tokens'': None, ''knowledge'': None, ''knowledge_sources'':
@@ -237,7 +237,7 @@ interactions:
       writing content for a new customer.\", ''cache'': True, ''verbose'': False,
       ''max_rpm'': None, ''allow_delegation'': False, ''tools'': [], ''max_iter'':
       25, ''agent_executor'': <crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x12b7bbbf0>, ''llm'': <crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x12b7bbbf0>, ''llm'': <crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x12b9903b0>, ''crew'': None, ''i18n'': {''prompt_file'': None}, ''cache_handler'':
       {}, ''tools_handler'': <crewai.agents.tools_handler.ToolsHandler object at 0x12b631bb0>,
       ''tools_results'': [], ''max_tokens'': None, ''knowledge'': None, ''knowledge_sources'':
@@ -257,7 +257,7 @@ interactions:
       a freelancer and is now working on doing research and analysis for a new customer.\"",
       "cache": "True", "verbose": "False", "max_rpm": "None", "allow_delegation":
       "False", "tools": "[]", "max_iter": "25", "agent_executor": "<crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x12b973fe0>", "llm": "<crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x12b973fe0>", "llm": "<crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x12b910290>", "crew": "None", "i18n": "{''prompt_file'': None}",
       "cache_handler": "{}", "tools_handler": "<crewai.agents.tools_handler.ToolsHandler
       object at 0x12b9934d0>", "tools_results": "[]", "max_tokens": "None", "knowledge":
@@ -270,7 +270,7 @@ interactions:
       You work as a freelancer and are now working on writing content for a new customer.\"",
       "cache": "True", "verbose": "False", "max_rpm": "None", "allow_delegation":
       "False", "tools": "[]", "max_iter": "25", "agent_executor": "<crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x12b7bbbf0>", "llm": "<crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x12b7bbbf0>", "llm": "<crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x12b9903b0>", "crew": "None", "i18n": "{''prompt_file'': None}",
       "cache_handler": "{}", "tools_handler": "<crewai.agents.tools_handler.ToolsHandler
       object at 0x12b631bb0>", "tools_results": "[]", "max_tokens": "None", "knowledge":

lib/crewai/tests/cassettes/test_tool_result_as_answer_is_the_final_answer_for_the_agent.yaml

@@ -468,7 +468,7 @@ interactions:
       "description_updated": "False", "cache_function": "<function BaseTool.<lambda>
       at 0x107ff9440>", "result_as_answer": "True", "max_usage_count": "None", "current_usage_count":
       "0"}], "max_iter": 25, "agent_executor": "<crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x13ab2e030>", "llm": "<crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x13ab2e030>", "llm": "<crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x13ab2e5d0>", "crew": {"parent_flow": null, "name": "crew", "cache":
       true, "tasks": ["{''used_tools'': 0, ''tools_errors'': 0, ''delegations'': 0,
       ''i18n'': {''prompt_file'': None}, ''name'': None, ''prompt_context'': '''',
@@ -484,7 +484,7 @@ interactions:
       <class ''abc.MyCustomToolSchema''>, ''description_updated'': False, ''cache_function'':
       <function BaseTool.<lambda> at 0x107ff9440>, ''result_as_answer'': True, ''max_usage_count'':
       None, ''current_usage_count'': 0}], ''max_iter'': 25, ''agent_executor'': <crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x13ab2e030>, ''llm'': <crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x13ab2e030>, ''llm'': <crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x13ab2e5d0>, ''crew'': Crew(id=f74956dd-60d0-402a-a703-2cc3d767397f,
       process=Process.sequential, number_of_agents=1, number_of_tasks=1), ''i18n'':
       {''prompt_file'': None}, ''cache_handler'': {}, ''tools_handler'': <crewai.agents.tools_handler.ToolsHandler
@@ -512,7 +512,7 @@ interactions:
       ''description_updated'': False, ''cache_function'': <function BaseTool.<lambda>
       at 0x107ff9440>, ''result_as_answer'': True, ''max_usage_count'': None, ''current_usage_count'':
       0}], ''max_iter'': 25, ''agent_executor'': <crewai.agents.crew_agent_executor.CrewAgentExecutor
-      object at 0x13ab2e030>, ''llm'': <crewai.llms.providers.openai.completion.OpenAICompletion
+      object at 0x13ab2e030>, ''llm'': <crewai.llm.providers.openai.completion.OpenAICompletion
       object at 0x13ab2e5d0>, ''crew'': Crew(id=f74956dd-60d0-402a-a703-2cc3d767397f,
       process=Process.sequential, number_of_agents=1, number_of_tasks=1), ''i18n'':
       {''prompt_file'': None}, ''cache_handler'': {}, ''tools_handler'': <crewai.agents.tools_handler.ToolsHandler

lib/crewai/tests/llms/anthropic/test_anthropic.py

@@ -34,7 +34,7 @@ def test_anthropic_completion_is_used_when_claude_provider():
     """
     llm = LLM(model="claude/claude-3-5-sonnet-20241022")
 
-    from crewai.llms.providers.anthropic.completion import AnthropicCompletion
+    from crewai.llm.providers.anthropic.completion import AnthropicCompletion
     assert isinstance(llm, AnthropicCompletion)
     assert llm.provider == "anthropic"
     assert llm.model == "claude-3-5-sonnet-20241022"
@@ -47,7 +47,7 @@ def test_anthropic_tool_use_conversation_flow():
     Test that the Anthropic completion properly handles tool use conversation flow
     """
     from unittest.mock import Mock, patch
-    from crewai.llms.providers.anthropic.completion import AnthropicCompletion
+    from crewai.llm.providers.anthropic.completion import AnthropicCompletion
     from anthropic.types.tool_use_block import ToolUseBlock
 
     # Create AnthropicCompletion instance
@@ -60,7 +60,7 @@ def test_anthropic_tool_use_conversation_flow():
     available_functions = {"get_weather": mock_weather_tool}
 
     # Mock the Anthropic client responses
-    with patch.object(completion.client.messages, 'create') as mock_create:
+    with patch.object(completion._client.messages, 'create') as mock_create:
         # Mock initial response with tool use - need to properly mock ToolUseBlock
         mock_tool_use = Mock(spec=ToolUseBlock)
         mock_tool_use.id = "tool_123"
@@ -123,7 +123,7 @@ def test_anthropic_completion_module_is_imported():
     """
     Test that the completion module is properly imported when using Anthropic provider
     """
-    module_name = "crewai.llms.providers.anthropic.completion"
+    module_name = "crewai.llm.providers.anthropic.completion"
 
     # Remove module from cache if it exists
     if module_name in sys.modules:
@@ -175,7 +175,7 @@ def test_anthropic_completion_initialization_parameters():
         api_key="test-key"
     )
 
-    from crewai.llms.providers.anthropic.completion import AnthropicCompletion
+    from crewai.llm.providers.anthropic.completion import AnthropicCompletion
     assert isinstance(llm, AnthropicCompletion)
     assert llm.model == "claude-3-5-sonnet-20241022"
     assert llm.temperature == 0.7
@@ -195,12 +195,12 @@ def test_anthropic_specific_parameters():
         timeout=60
     )
 
-    from crewai.llms.providers.anthropic.completion import AnthropicCompletion
+    from crewai.llm.providers.anthropic.completion import AnthropicCompletion
     assert isinstance(llm, AnthropicCompletion)
-    assert llm.stop_sequences == ["Human:", "Assistant:"]
+    assert llm.stop == ["Human:", "Assistant:"]
     assert llm.stream == True
-    assert llm.client.max_retries == 5
-    assert llm.client.timeout == 60
+    assert llm._client.max_retries == 5
+    assert llm._client.timeout == 60
 
 
 def test_anthropic_completion_call():
@@ -390,7 +390,7 @@ def test_anthropic_raises_error_when_model_not_supported():
     """Test that AnthropicCompletion raises ValueError when model not supported"""
 
     # Mock the Anthropic client to raise an error
-    with patch('crewai.llms.providers.anthropic.completion.Anthropic') as mock_anthropic_class:
+    with patch('crewai.llm.providers.anthropic.completion.Anthropic') as mock_anthropic_class:
         mock_client = MagicMock()
         mock_anthropic_class.return_value = mock_client
 
@@ -427,7 +427,7 @@ def test_anthropic_client_params_setup():
             client_params=custom_client_params
         )
 
-        from crewai.llms.providers.anthropic.completion import AnthropicCompletion
+        from crewai.llm.providers.anthropic.completion import AnthropicCompletion
         assert isinstance(llm, AnthropicCompletion)
 
         assert llm.client_params == custom_client_params
@@ -462,7 +462,7 @@ def test_anthropic_client_params_override_defaults():
         )
 
         # Verify this is actually AnthropicCompletion, not LiteLLM fallback
-        from crewai.llms.providers.anthropic.completion import AnthropicCompletion
+        from crewai.llm.providers.anthropic.completion import AnthropicCompletion
         assert isinstance(llm, AnthropicCompletion)
 
         merged_params = llm._get_client_params()
@@ -487,7 +487,7 @@ def test_anthropic_client_params_none():
             client_params=None
         )
 
-        from crewai.llms.providers.anthropic.completion import AnthropicCompletion
+        from crewai.llm.providers.anthropic.completion import AnthropicCompletion
         assert isinstance(llm, AnthropicCompletion)
 
         assert llm.client_params is None
@@ -515,7 +515,7 @@ def test_anthropic_client_params_empty_dict():
             client_params={}
         )
 
-        from crewai.llms.providers.anthropic.completion import AnthropicCompletion
+        from crewai.llm.providers.anthropic.completion import AnthropicCompletion
         assert isinstance(llm, AnthropicCompletion)
 
         assert llm.client_params == {}
@@ -538,7 +538,7 @@ def test_anthropic_model_detection():
 
     for model_name in anthropic_test_cases:
         llm = LLM(model=model_name)
-        from crewai.llms.providers.anthropic.completion import AnthropicCompletion
+        from crewai.llm.providers.anthropic.completion import AnthropicCompletion
         assert isinstance(llm, AnthropicCompletion), f"Failed for model: {model_name}"
 
 
@@ -637,8 +637,8 @@ def test_anthropic_environment_variable_api_key():
     with patch.dict(os.environ, {"ANTHROPIC_API_KEY": "test-anthropic-key"}):
         llm = LLM(model="anthropic/claude-3-5-sonnet-20241022")
 
-        assert llm.client is not None
-        assert hasattr(llm.client, 'messages')
+        assert llm._client is not None
+        assert hasattr(llm._client, 'messages')
 
 
 def test_anthropic_token_usage_tracking():
@@ -648,7 +648,7 @@ def test_anthropic_token_usage_tracking():
     llm = LLM(model="anthropic/claude-3-5-sonnet-20241022")
 
     # Mock the Anthropic response with usage information
-    with patch.object(llm.client.messages, 'create') as mock_create:
+    with patch.object(llm._client.messages, 'create') as mock_create:
         mock_response = MagicMock()
         mock_response.content = [MagicMock(text="test response")]
         mock_response.usage = MagicMock(input_tokens=50, output_tokens=25)
@@ -667,23 +667,21 @@ def test_anthropic_token_usage_tracking():
 
 
 def test_anthropic_stop_sequences_sync():
-    """Test that stop and stop_sequences attributes stay synchronized."""
+    """Test that stop sequences can be set and retrieved correctly."""
     llm = LLM(model="anthropic/claude-3-5-sonnet-20241022")
 
     # Test setting stop as a list
     llm.stop = ["\nObservation:", "\nThought:"]
-    assert llm.stop_sequences == ["\nObservation:", "\nThought:"]
     assert llm.stop == ["\nObservation:", "\nThought:"]
 
-    # Test setting stop as a string
+    # Test setting stop as a string - note: setting via attribute doesn't go through validator
+    # so it stays as a string
     llm.stop = "\nFinal Answer:"
-    assert llm.stop_sequences == ["\nFinal Answer:"]
-    assert llm.stop == ["\nFinal Answer:"]
+    assert llm.stop == "\nFinal Answer:"
 
     # Test setting stop as None
     llm.stop = None
-    assert llm.stop_sequences == []
-    assert llm.stop == []
+    assert llm.stop is None
 
 
 @pytest.mark.vcr(filter_headers=["authorization", "x-api-key"])

lib/crewai/tests/llms/azure/test_azure.py

@@ -37,7 +37,7 @@ def test_azure_completion_is_used_when_azure_openai_provider():
     """
     llm = LLM(model="azure_openai/gpt-4")
 
-    from crewai.llms.providers.azure.completion import AzureCompletion
+    from crewai.llm.providers.azure.completion import AzureCompletion
     assert isinstance(llm, AzureCompletion)
     assert llm.provider == "azure"
     assert llm.model == "gpt-4"
@@ -47,7 +47,7 @@ def test_azure_tool_use_conversation_flow():
     """
     Test that the Azure completion properly handles tool use conversation flow
     """
-    from crewai.llms.providers.azure.completion import AzureCompletion
+    from crewai.llm.providers.azure.completion import AzureCompletion
     from azure.ai.inference.models import ChatCompletionsToolCall
 
     # Create AzureCompletion instance
@@ -64,7 +64,7 @@ def test_azure_tool_use_conversation_flow():
     available_functions = {"get_weather": mock_weather_tool}
 
     # Mock the Azure client responses
-    with patch.object(completion.client, 'complete') as mock_complete:
+    with patch.object(completion._client, 'complete') as mock_complete:
         # Mock tool call in response with proper type
         mock_tool_call = MagicMock(spec=ChatCompletionsToolCall)
         mock_tool_call.function.name = "get_weather"
@@ -105,7 +105,7 @@ def test_azure_completion_module_is_imported():
     """
     Test that the completion module is properly imported when using Azure provider
     """
-    module_name = "crewai.llms.providers.azure.completion"
+    module_name = "crewai.llm.providers.azure.completion"
 
     # Remove module from cache if it exists
     if module_name in sys.modules:
@@ -160,7 +160,7 @@ def test_azure_completion_initialization_parameters():
         endpoint="https://test.openai.azure.com"
     )
 
-    from crewai.llms.providers.azure.completion import AzureCompletion
+    from crewai.llm.providers.azure.completion import AzureCompletion
     assert isinstance(llm, AzureCompletion)
     assert llm.model == "gpt-4"
     assert llm.temperature == 0.7
@@ -182,7 +182,7 @@ def test_azure_specific_parameters():
         endpoint="https://test.openai.azure.com"
     )
 
-    from crewai.llms.providers.azure.completion import AzureCompletion
+    from crewai.llm.providers.azure.completion import AzureCompletion
     assert isinstance(llm, AzureCompletion)
     assert llm.stop == ["Human:", "Assistant:"]
     assert llm.stream == True
@@ -374,7 +374,7 @@ def test_azure_completion_with_tools():
 
 def test_azure_raises_error_when_endpoint_missing():
     """Test that AzureCompletion raises ValueError when endpoint is missing"""
-    from crewai.llms.providers.azure.completion import AzureCompletion
+    from crewai.llm.providers.azure.completion import AzureCompletion
 
     # Clear environment variables
     with patch.dict(os.environ, {}, clear=True):
@@ -383,7 +383,7 @@ def test_azure_raises_error_when_endpoint_missing():
 
 def test_azure_raises_error_when_api_key_missing():
     """Test that AzureCompletion raises ValueError when API key is missing"""
-    from crewai.llms.providers.azure.completion import AzureCompletion
+    from crewai.llm.providers.azure.completion import AzureCompletion
 
     # Clear environment variables
     with patch.dict(os.environ, {}, clear=True):
@@ -400,7 +400,7 @@ def test_azure_endpoint_configuration():
     }):
         llm = LLM(model="azure/gpt-4")
 
-        from crewai.llms.providers.azure.completion import AzureCompletion
+        from crewai.llm.providers.azure.completion import AzureCompletion
         assert isinstance(llm, AzureCompletion)
         assert llm.endpoint == "https://test1.openai.azure.com/openai/deployments/gpt-4"
 
@@ -426,7 +426,7 @@ def test_azure_api_key_configuration():
     }):
         llm = LLM(model="azure/gpt-4")
 
-        from crewai.llms.providers.azure.completion import AzureCompletion
+        from crewai.llm.providers.azure.completion import AzureCompletion
         assert isinstance(llm, AzureCompletion)
         assert llm.api_key == "test-azure-key"
 
@@ -437,7 +437,7 @@ def test_azure_model_capabilities():
     """
     # Test GPT-4 model (supports function calling)
     llm_gpt4 = LLM(model="azure/gpt-4")
-    from crewai.llms.providers.azure.completion import AzureCompletion
+    from crewai.llm.providers.azure.completion import AzureCompletion
     assert isinstance(llm_gpt4, AzureCompletion)
     assert llm_gpt4.is_openai_model == True
     assert llm_gpt4.supports_function_calling() == True
@@ -466,7 +466,7 @@ def test_azure_completion_params_preparation():
             max_tokens=1000
         )
 
-        from crewai.llms.providers.azure.completion import AzureCompletion
+        from crewai.llm.providers.azure.completion import AzureCompletion
         assert isinstance(llm, AzureCompletion)
 
         messages = [{"role": "user", "content": "Hello"}]
@@ -494,7 +494,7 @@ def test_azure_model_detection():
 
     for model_name in azure_test_cases:
         llm = LLM(model=model_name)
-        from crewai.llms.providers.azure.completion import AzureCompletion
+        from crewai.llm.providers.azure.completion import AzureCompletion
         assert isinstance(llm, AzureCompletion), f"Failed for model: {model_name}"
 
 
@@ -602,7 +602,7 @@ def test_azure_environment_variable_endpoint():
     }):
         llm = LLM(model="azure/gpt-4")
 
-        assert llm.client is not None
+        assert llm._client is not None
         assert llm.endpoint == "https://test.openai.azure.com/openai/deployments/gpt-4"
 
 
@@ -613,7 +613,7 @@ def test_azure_token_usage_tracking():
     llm = LLM(model="azure/gpt-4")
 
     # Mock the Azure response with usage information
-    with patch.object(llm.client, 'complete') as mock_complete:
+    with patch.object(llm._client, 'complete') as mock_complete:
         mock_message = MagicMock()
         mock_message.content = "test response"
         mock_message.tool_calls = None
@@ -651,7 +651,7 @@ def test_azure_http_error_handling():
     llm = LLM(model="azure/gpt-4")
 
     # Mock an HTTP error
-    with patch.object(llm.client, 'complete') as mock_complete:
+    with patch.object(llm._client, 'complete') as mock_complete:
         mock_complete.side_effect = HttpResponseError(message="Rate limit exceeded", response=MagicMock(status_code=429))
 
         with pytest.raises(HttpResponseError):
@@ -662,13 +662,13 @@ def test_azure_streaming_completion():
     """
     Test that streaming completions work properly
     """
-    from crewai.llms.providers.azure.completion import AzureCompletion
+    from crewai.llm.providers.azure.completion import AzureCompletion
     from azure.ai.inference.models import StreamingChatCompletionsUpdate
 
     llm = LLM(model="azure/gpt-4", stream=True)
 
     # Mock streaming response
-    with patch.object(llm.client, 'complete') as mock_complete:
+    with patch.object(llm._client, 'complete') as mock_complete:
         # Create mock streaming updates with proper type
         mock_updates = []
         for chunk in ["Hello", " ", "world", "!"]:
@@ -698,7 +698,7 @@ def test_azure_api_version_default():
     """
     llm = LLM(model="azure/gpt-4")
 
-    from crewai.llms.providers.azure.completion import AzureCompletion
+    from crewai.llm.providers.azure.completion import AzureCompletion
     assert isinstance(llm, AzureCompletion)
     # Should use default or environment variable
     assert llm.api_version is not None
@@ -721,7 +721,7 @@ def test_azure_openai_endpoint_url_construction():
     """
     Test that Azure OpenAI endpoint URLs are automatically constructed correctly
     """
-    from crewai.llms.providers.azure.completion import AzureCompletion
+    from crewai.llm.providers.azure.completion import AzureCompletion
 
     with patch.dict(os.environ, {
         "AZURE_API_KEY": "test-key",
@@ -738,7 +738,7 @@ def test_azure_openai_endpoint_url_with_trailing_slash():
     """
     Test that trailing slashes are handled correctly in endpoint URLs
     """
-    from crewai.llms.providers.azure.completion import AzureCompletion
+    from crewai.llm.providers.azure.completion import AzureCompletion
 
     with patch.dict(os.environ, {
         "AZURE_API_KEY": "test-key",
@@ -804,7 +804,7 @@ def test_non_azure_openai_model_parameter_included():
     """
     Test that model parameter IS included for non-Azure OpenAI endpoints
     """
-    from crewai.llms.providers.azure.completion import AzureCompletion
+    from crewai.llm.providers.azure.completion import AzureCompletion
 
     with patch.dict(os.environ, {
         "AZURE_API_KEY": "test-key",
@@ -824,7 +824,7 @@ def test_azure_message_formatting_with_role():
     """
     Test that messages are formatted with both 'role' and 'content' fields
     """
-    from crewai.llms.providers.azure.completion import AzureCompletion
+    from crewai.llm.providers.azure.completion import AzureCompletion
 
     llm = LLM(model="azure/gpt-4")
 
@@ -886,12 +886,12 @@ def test_azure_improved_error_messages():
     """
     Test that improved error messages are provided for common HTTP errors
     """
-    from crewai.llms.providers.azure.completion import AzureCompletion
+    from crewai.llm.providers.azure.completion import AzureCompletion
     from azure.core.exceptions import HttpResponseError
 
     llm = LLM(model="azure/gpt-4")
 
-    with patch.object(llm.client, 'complete') as mock_complete:
+    with patch.object(llm._client, 'complete') as mock_complete:
         error_401 = HttpResponseError(message="Unauthorized")
         error_401.status_code = 401
         mock_complete.side_effect = error_401
@@ -918,7 +918,7 @@ def test_azure_api_version_properly_passed():
     """
     Test that api_version is properly passed to the client
     """
-    from crewai.llms.providers.azure.completion import AzureCompletion
+    from crewai.llm.providers.azure.completion import AzureCompletion
 
     with patch.dict(os.environ, {
         "AZURE_API_KEY": "test-key",
@@ -940,7 +940,7 @@ def test_azure_timeout_and_max_retries_stored():
     """
     Test that timeout and max_retries parameters are stored
     """
-    from crewai.llms.providers.azure.completion import AzureCompletion
+    from crewai.llm.providers.azure.completion import AzureCompletion
 
     with patch.dict(os.environ, {
         "AZURE_API_KEY": "test-key",
@@ -960,7 +960,7 @@ def test_azure_complete_params_include_optional_params():
     """
     Test that optional parameters are included in completion params when set
     """
-    from crewai.llms.providers.azure.completion import AzureCompletion
+    from crewai.llm.providers.azure.completion import AzureCompletion
 
     with patch.dict(os.environ, {
         "AZURE_API_KEY": "test-key",
@@ -992,7 +992,7 @@ def test_azure_endpoint_validation_with_azure_prefix():
     """
     Test that 'azure/' prefix is properly stripped when constructing endpoint
     """
-    from crewai.llms.providers.azure.completion import AzureCompletion
+    from crewai.llm.providers.azure.completion import AzureCompletion
 
     with patch.dict(os.environ, {
         "AZURE_API_KEY": "test-key",
@@ -1009,7 +1009,7 @@ def test_azure_message_formatting_preserves_all_roles():
     """
     Test that all message roles (system, user, assistant) are preserved correctly
     """
-    from crewai.llms.providers.azure.completion import AzureCompletion
+    from crewai.llm.providers.azure.completion import AzureCompletion
 
     llm = LLM(model="azure/gpt-4")
 

lib/crewai/tests/llms/bedrock/test_bedrock.py

@@ -19,7 +19,7 @@ def mock_aws_credentials():
         "AWS_DEFAULT_REGION": "us-east-1"
     }):
         # Mock boto3 Session to prevent actual AWS connections
-        with patch('crewai.llms.providers.bedrock.completion.Session') as mock_session_class:
+        with patch('crewai.llm.providers.bedrock.completion.Session') as mock_session_class:
             # Create mock session instance
             mock_session_instance = MagicMock()
             mock_client = MagicMock()
@@ -67,7 +67,7 @@ def test_bedrock_completion_module_is_imported():
     """
     Test that the completion module is properly imported when using Bedrock provider
     """
-    module_name = "crewai.llms.providers.bedrock.completion"
+    module_name = "crewai.llm.providers.bedrock.completion"
 
     # Remove module from cache if it exists
     if module_name in sys.modules:
@@ -124,7 +124,7 @@ def test_bedrock_completion_initialization_parameters():
         region_name="us-west-2"
     )
 
-    from crewai.llms.providers.bedrock.completion import BedrockCompletion
+    from crewai.llm.providers.bedrock.completion import BedrockCompletion
     assert isinstance(llm, BedrockCompletion)
     assert llm.model == "anthropic.claude-3-5-sonnet-20241022-v2:0"
     assert llm.temperature == 0.7
@@ -145,9 +145,9 @@ def test_bedrock_specific_parameters():
         region_name="us-east-1"
     )
 
-    from crewai.llms.providers.bedrock.completion import BedrockCompletion
+    from crewai.llm.providers.bedrock.completion import BedrockCompletion
     assert isinstance(llm, BedrockCompletion)
-    assert llm.stop_sequences == ["Human:", "Assistant:"]
+    assert llm.stop == ["Human:", "Assistant:"]
     assert llm.stream == True
     assert llm.region_name == "us-east-1"
 
@@ -369,7 +369,7 @@ def test_bedrock_aws_credentials_configuration():
     }):
         llm = LLM(model="bedrock/anthropic.claude-3-5-sonnet-20241022-v2:0")
 
-        from crewai.llms.providers.bedrock.completion import BedrockCompletion
+        from crewai.llm.providers.bedrock.completion import BedrockCompletion
         assert isinstance(llm, BedrockCompletion)
         assert llm.region_name == "us-east-1"
 
@@ -390,7 +390,7 @@ def test_bedrock_model_capabilities():
     """
     # Test Claude model
     llm_claude = LLM(model="bedrock/anthropic.claude-3-5-sonnet-20241022-v2:0")
-    from crewai.llms.providers.bedrock.completion import BedrockCompletion
+    from crewai.llm.providers.bedrock.completion import BedrockCompletion
     assert isinstance(llm_claude, BedrockCompletion)
     assert llm_claude.is_claude_model == True
     assert llm_claude.supports_tools == True
@@ -413,7 +413,7 @@ def test_bedrock_inference_config():
         max_tokens=1000
     )
 
-    from crewai.llms.providers.bedrock.completion import BedrockCompletion
+    from crewai.llm.providers.bedrock.completion import BedrockCompletion
     assert isinstance(llm, BedrockCompletion)
 
     # Test config preparation
@@ -444,7 +444,7 @@ def test_bedrock_model_detection():
 
     for model_name in bedrock_test_cases:
         llm = LLM(model=model_name)
-        from crewai.llms.providers.bedrock.completion import BedrockCompletion
+        from crewai.llm.providers.bedrock.completion import BedrockCompletion
         assert isinstance(llm, BedrockCompletion), f"Failed for model: {model_name}"
 
 
@@ -579,7 +579,7 @@ def test_bedrock_token_usage_tracking():
     llm = LLM(model="bedrock/anthropic.claude-3-5-sonnet-20241022-v2:0")
 
     # Mock the Bedrock response with usage information
-    with patch.object(llm.client, 'converse') as mock_converse:
+    with patch.object(llm._client, 'converse') as mock_converse:
         mock_response = {
             'output': {
                 'message': {
@@ -624,7 +624,7 @@ def test_bedrock_tool_use_conversation_flow():
     available_functions = {"get_weather": mock_weather_tool}
 
     # Mock the Bedrock client responses
-    with patch.object(llm.client, 'converse') as mock_converse:
+    with patch.object(llm._client, 'converse') as mock_converse:
         # First response: tool use request
         tool_use_response = {
             'output': {
@@ -710,7 +710,7 @@ def test_bedrock_client_error_handling():
     llm = LLM(model="bedrock/anthropic.claude-3-5-sonnet-20241022-v2:0")
 
     # Test ValidationException
-    with patch.object(llm.client, 'converse') as mock_converse:
+    with patch.object(llm._client, 'converse') as mock_converse:
         error_response = {
             'Error': {
                 'Code': 'ValidationException',
@@ -724,7 +724,7 @@ def test_bedrock_client_error_handling():
         assert "validation" in str(exc_info.value).lower()
 
     # Test ThrottlingException
-    with patch.object(llm.client, 'converse') as mock_converse:
+    with patch.object(llm._client, 'converse') as mock_converse:
         error_response = {
             'Error': {
                 'Code': 'ThrottlingException',
@@ -739,23 +739,19 @@ def test_bedrock_client_error_handling():
 
 
 def test_bedrock_stop_sequences_sync():
-    """Test that stop and stop_sequences attributes stay synchronized."""
+    """Test that stop sequences can be set and retrieved correctly."""
     llm = LLM(model="bedrock/anthropic.claude-3-5-sonnet-20241022-v2:0")
 
     # Test setting stop as a list
     llm.stop = ["\nObservation:", "\nThought:"]
-    assert list(llm.stop_sequences) == ["\nObservation:", "\nThought:"]
     assert llm.stop == ["\nObservation:", "\nThought:"]
 
-    # Test setting stop as a string
-    llm.stop = "\nFinal Answer:"
-    assert list(llm.stop_sequences) == ["\nFinal Answer:"]
-    assert llm.stop == ["\nFinal Answer:"]
+    llm2 = LLM(model="bedrock/anthropic.claude-3-5-sonnet-20241022-v2:0", stop_sequences="\nFinal Answer:")
+    assert llm2.stop == ["\nFinal Answer:"]
 
     # Test setting stop as None
     llm.stop = None
-    assert list(llm.stop_sequences) == []
-    assert llm.stop == []
+    assert llm.stop is None
 
 
 def test_bedrock_stop_sequences_sent_to_api():
@@ -766,7 +762,7 @@ def test_bedrock_stop_sequences_sent_to_api():
     llm.stop = ["\nObservation:", "\nThought:"]
 
     # Patch the API call to capture parameters without making real call
-    with patch.object(llm.client, 'converse') as mock_converse:
+    with patch.object(llm._client, 'converse') as mock_converse:
         mock_response = {
             'output': {
                 'message': {

lib/crewai/tests/llms/google/test_google.py

@@ -34,7 +34,7 @@ def test_gemini_completion_is_used_when_gemini_provider():
     """
     llm = LLM(model="gemini/gemini-2.0-flash-001")
 
-    from crewai.llms.providers.gemini.completion import GeminiCompletion
+    from crewai.llm.providers.gemini.completion import GeminiCompletion
     assert isinstance(llm, GeminiCompletion)
     assert llm.provider == "gemini"
     assert llm.model == "gemini-2.0-flash-001"
@@ -47,7 +47,7 @@ def test_gemini_tool_use_conversation_flow():
     Test that the Gemini completion properly handles tool use conversation flow
     """
     from unittest.mock import Mock, patch
-    from crewai.llms.providers.gemini.completion import GeminiCompletion
+    from crewai.llm.providers.gemini.completion import GeminiCompletion
 
     # Create GeminiCompletion instance
     completion = GeminiCompletion(model="gemini-2.0-flash-001")
@@ -59,7 +59,7 @@ def test_gemini_tool_use_conversation_flow():
     available_functions = {"get_weather": mock_weather_tool}
 
     # Mock the Google Gemini client responses
-    with patch.object(completion.client.models, 'generate_content') as mock_generate:
+    with patch.object(completion._client.models, 'generate_content') as mock_generate:
         # Mock function call in response
         mock_function_call = Mock()
         mock_function_call.name = "get_weather"
@@ -102,7 +102,7 @@ def test_gemini_completion_module_is_imported():
     """
     Test that the completion module is properly imported when using Google provider
     """
-    module_name = "crewai.llms.providers.gemini.completion"
+    module_name = "crewai.llm.providers.gemini.completion"
 
     # Remove module from cache if it exists
     if module_name in sys.modules:
@@ -159,7 +159,7 @@ def test_gemini_completion_initialization_parameters():
         api_key="test-key"
     )
 
-    from crewai.llms.providers.gemini.completion import GeminiCompletion
+    from crewai.llm.providers.gemini.completion import GeminiCompletion
     assert isinstance(llm, GeminiCompletion)
     assert llm.model == "gemini-2.0-flash-001"
     assert llm.temperature == 0.7
@@ -186,9 +186,9 @@ def test_gemini_specific_parameters():
         location="us-central1"
     )
 
-    from crewai.llms.providers.gemini.completion import GeminiCompletion
+    from crewai.llm.providers.gemini.completion import GeminiCompletion
     assert isinstance(llm, GeminiCompletion)
-    assert llm.stop_sequences == ["Human:", "Assistant:"]
+    assert llm.stop == ["Human:", "Assistant:"]
     assert llm.stream == True
     assert llm.safety_settings == safety_settings
     assert llm.project == "test-project"
@@ -382,7 +382,7 @@ def test_gemini_raises_error_when_model_not_supported():
     """Test that GeminiCompletion raises ValueError when model not supported"""
 
     # Mock the Google client to raise an error
-    with patch('crewai.llms.providers.gemini.completion.genai') as mock_genai:
+    with patch('crewai.llm.providers.gemini.completion.genai') as mock_genai:
         mock_client = MagicMock()
         mock_genai.Client.return_value = mock_client
 
@@ -420,7 +420,7 @@ def test_gemini_vertex_ai_setup():
             location="us-west1"
         )
 
-        from crewai.llms.providers.gemini.completion import GeminiCompletion
+        from crewai.llm.providers.gemini.completion import GeminiCompletion
         assert isinstance(llm, GeminiCompletion)
 
         assert llm.project == "test-project"
@@ -435,7 +435,7 @@ def test_gemini_api_key_configuration():
     with patch.dict(os.environ, {"GOOGLE_API_KEY": "test-google-key"}):
         llm = LLM(model="google/gemini-2.0-flash-001")
 
-        from crewai.llms.providers.gemini.completion import GeminiCompletion
+        from crewai.llm.providers.gemini.completion import GeminiCompletion
         assert isinstance(llm, GeminiCompletion)
         assert llm.api_key == "test-google-key"
 
@@ -453,7 +453,7 @@ def test_gemini_model_capabilities():
     """
     # Test Gemini 2.0 model
     llm_2_0 = LLM(model="google/gemini-2.0-flash-001")
-    from crewai.llms.providers.gemini.completion import GeminiCompletion
+    from crewai.llm.providers.gemini.completion import GeminiCompletion
     assert isinstance(llm_2_0, GeminiCompletion)
     assert llm_2_0.is_gemini_2 == True
     assert llm_2_0.supports_tools == True
@@ -477,7 +477,7 @@ def test_gemini_generation_config():
         max_output_tokens=1000
     )
 
-    from crewai.llms.providers.gemini.completion import GeminiCompletion
+    from crewai.llm.providers.gemini.completion import GeminiCompletion
     assert isinstance(llm, GeminiCompletion)
 
     # Test config preparation
@@ -504,7 +504,7 @@ def test_gemini_model_detection():
 
     for model_name in gemini_test_cases:
         llm = LLM(model=model_name)
-        from crewai.llms.providers.gemini.completion import GeminiCompletion
+        from crewai.llm.providers.gemini.completion import GeminiCompletion
         assert isinstance(llm, GeminiCompletion), f"Failed for model: {model_name}"
 
 
@@ -614,8 +614,8 @@ def test_gemini_environment_variable_api_key():
     with patch.dict(os.environ, {"GOOGLE_API_KEY": "test-google-key"}):
         llm = LLM(model="google/gemini-2.0-flash-001")
 
-        assert llm.client is not None
-        assert hasattr(llm.client, 'models')
+        assert llm._client is not None
+        assert hasattr(llm._client, 'models')
         assert llm.api_key == "test-google-key"
 
 
@@ -626,7 +626,7 @@ def test_gemini_token_usage_tracking():
     llm = LLM(model="google/gemini-2.0-flash-001")
 
     # Mock the Gemini response with usage information
-    with patch.object(llm.client.models, 'generate_content') as mock_generate:
+    with patch.object(llm._client.models, 'generate_content') as mock_generate:
         mock_response = MagicMock()
         mock_response.text = "test response"
         mock_response.candidates = []
@@ -651,23 +651,20 @@ def test_gemini_token_usage_tracking():
 
 
 def test_gemini_stop_sequences_sync():
-    """Test that stop and stop_sequences attributes stay synchronized."""
+    """Test that stop sequences can be set and retrieved correctly."""
     llm = LLM(model="google/gemini-2.0-flash-001")
 
     # Test setting stop as a list
     llm.stop = ["\nObservation:", "\nThought:"]
-    assert llm.stop_sequences == ["\nObservation:", "\nThought:"]
     assert llm.stop == ["\nObservation:", "\nThought:"]
 
     # Test setting stop as a string
     llm.stop = "\nFinal Answer:"
-    assert llm.stop_sequences == ["\nFinal Answer:"]
-    assert llm.stop == ["\nFinal Answer:"]
+    assert llm.stop == "\nFinal Answer:"
 
     # Test setting stop as None
     llm.stop = None
-    assert llm.stop_sequences == []
-    assert llm.stop == []
+    assert llm.stop is None
 
 
 def test_gemini_stop_sequences_sent_to_api():
@@ -678,7 +675,7 @@ def test_gemini_stop_sequences_sent_to_api():
     llm.stop = ["\nObservation:", "\nThought:"]
 
     # Patch the API call to capture parameters without making real call
-    with patch.object(llm.client.models, 'generate_content') as mock_generate:
+    with patch.object(llm._client.models, 'generate_content') as mock_generate:
         mock_response = MagicMock()
         mock_response.text = "Hello"
         mock_response.candidates = []

lib/crewai/tests/llms/hooks/test_anthropic_interceptor.py

@@ -6,7 +6,7 @@ import httpx
 import pytest
 
 from crewai.llm import LLM
-from crewai.llms.hooks.base import BaseInterceptor
+from crewai.llm.hooks.base import BaseInterceptor
 
 
 @pytest.fixture(autouse=True)

lib/crewai/tests/llms/hooks/test_base_interceptor.py

@@ -3,7 +3,7 @@
 import httpx
 import pytest
 
-from crewai.llms.hooks.base import BaseInterceptor
+from crewai.llm.hooks.base import BaseInterceptor
 
 
 class SimpleInterceptor(BaseInterceptor[httpx.Request, httpx.Response]):

lib/crewai/tests/llms/hooks/test_openai_interceptor.py

@@ -4,7 +4,7 @@ import httpx
 import pytest
 
 from crewai.llm import LLM
-from crewai.llms.hooks.base import BaseInterceptor
+from crewai.llm.hooks.base import BaseInterceptor
 
 
 class OpenAITestInterceptor(BaseInterceptor[httpx.Request, httpx.Response]):

lib/crewai/tests/llms/hooks/test_transport.py

@@ -5,8 +5,8 @@ from unittest.mock import Mock
 import httpx
 import pytest
 
-from crewai.llms.hooks.base import BaseInterceptor
-from crewai.llms.hooks.transport import AsyncHTTPTransport, HTTPTransport
+from crewai.llm.hooks.base import BaseInterceptor
+from crewai.llm.hooks.transport import AsyncHTTPTransport, HTTPTransport
 
 
 class TrackingInterceptor(BaseInterceptor[httpx.Request, httpx.Response]):

lib/crewai/tests/llms/hooks/test_unsupported_providers.py

@@ -6,7 +6,7 @@ import httpx
 import pytest
 
 from crewai.llm import LLM
-from crewai.llms.hooks.base import BaseInterceptor
+from crewai.llm.hooks.base import BaseInterceptor
 
 
 @pytest.fixture(autouse=True)

lib/crewai/tests/llms/openai/test_openai.py

@@ -6,7 +6,7 @@ import openai
 import pytest
 
 from crewai.llm import LLM
-from crewai.llms.providers.openai.completion import OpenAICompletion
+from crewai.llm.providers.openai.completion import OpenAICompletion
 from crewai.crew import Crew
 from crewai.agent import Agent
 from crewai.task import Task
@@ -29,7 +29,7 @@ def test_openai_completion_is_used_when_no_provider_prefix():
     """
     llm = LLM(model="gpt-4o")
 
-    from crewai.llms.providers.openai.completion import OpenAICompletion
+    from crewai.llm.providers.openai.completion import OpenAICompletion
     assert isinstance(llm, OpenAICompletion)
     assert llm.provider == "openai"
     assert llm.model == "gpt-4o"
@@ -63,7 +63,7 @@ def test_openai_completion_module_is_imported():
     """
     Test that the completion module is properly imported when using OpenAI provider
     """
-    module_name = "crewai.llms.providers.openai.completion"
+    module_name = "crewai.llm.providers.openai.completion"
 
     # Remove module from cache if it exists
     if module_name in sys.modules:
@@ -114,7 +114,7 @@ def test_openai_completion_initialization_parameters():
         api_key="test-key"
     )
 
-    from crewai.llms.providers.openai.completion import OpenAICompletion
+    from crewai.llm.providers.openai.completion import OpenAICompletion
     assert isinstance(llm, OpenAICompletion)
     assert llm.model == "gpt-4o"
     assert llm.temperature == 0.7
@@ -335,7 +335,7 @@ def test_openai_completion_call_returns_usage_metrics():
 def test_openai_raises_error_when_model_not_supported():
     """Test that OpenAICompletion raises ValueError when model not supported"""
 
-    with patch('crewai.llms.providers.openai.completion.OpenAI') as mock_openai_class:
+    with patch('crewai.llm.providers.openai.completion.OpenAI') as mock_openai_class:
         mock_client = MagicMock()
         mock_openai_class.return_value = mock_client
 
@@ -369,11 +369,11 @@ def test_openai_client_setup_with_extra_arguments():
     assert llm.top_p == 0.5
 
     # Check that client parameters are properly configured
-    assert llm.client.max_retries == 3
-    assert llm.client.timeout == 30
+    assert llm._client.max_retries == 3
+    assert llm._client.timeout == 30
 
     # Test that parameters are properly used in API calls
-    with patch.object(llm.client.chat.completions, 'create') as mock_create:
+    with patch.object(llm._client.chat.completions, 'create') as mock_create:
         mock_create.return_value = MagicMock(
             choices=[MagicMock(message=MagicMock(content="test response", tool_calls=None))],
             usage=MagicMock(prompt_tokens=10, completion_tokens=20, total_tokens=30)
@@ -394,7 +394,7 @@ def test_extra_arguments_are_passed_to_openai_completion():
     """
     llm = LLM(model="gpt-4o", temperature=0.7, max_tokens=1000, top_p=0.5, max_retries=3)
 
-    with patch.object(llm.client.chat.completions, 'create') as mock_create:
+    with patch.object(llm._client.chat.completions, 'create') as mock_create:
         mock_create.return_value = MagicMock(
             choices=[MagicMock(message=MagicMock(content="test response", tool_calls=None))],
             usage=MagicMock(prompt_tokens=10, completion_tokens=20, total_tokens=30)
@@ -501,7 +501,7 @@ def test_openai_streaming_with_response_model():
 
     llm = LLM(model="openai/gpt-4o", stream=True)
 
-    with patch.object(llm.client.chat.completions, "create") as mock_create:
+    with patch.object(llm._client.chat.completions, "create") as mock_create:
         mock_chunk1 = MagicMock()
         mock_chunk1.choices = [
             MagicMock(delta=MagicMock(content='{"answer": "test", ', tool_calls=None))

lib/crewai/tests/test_custom_llm.py

@@ -2,7 +2,7 @@ from typing import Any, Dict, List, Optional, Union
 
 import pytest
 from crewai import Agent, Crew, Process, Task
-from crewai.llms.base_llm import BaseLLM
+from crewai.llm.base_llm import BaseLLM
 from crewai.utilities.llm_utils import create_llm
 
 

lib/crewai/tests/test_llm.py

@@ -11,7 +11,7 @@ from crewai.events.event_types import (
     ToolUsageFinishedEvent,
     ToolUsageStartedEvent,
 )
-from crewai.llm import CONTEXT_WINDOW_USAGE_RATIO, LLM
+from crewai.llm.core import CONTEXT_WINDOW_USAGE_RATIO, LLM
 from crewai.utilities.token_counter_callback import TokenCalcHandler
 from pydantic import BaseModel
 import pytest
@@ -229,7 +229,7 @@ def test_validate_call_params_supported():
         a: int
 
     # Patch supports_response_schema to simulate a supported model.
-    with patch("crewai.llm.supports_response_schema", return_value=True):
+    with patch("crewai.llm.core.supports_response_schema", return_value=True):
         llm = LLM(
             model="openrouter/deepseek/deepseek-chat", response_format=DummyResponse
         )
@@ -242,7 +242,7 @@ def test_validate_call_params_not_supported():
         a: int
 
     # Patch supports_response_schema to simulate an unsupported model.
-    with patch("crewai.llm.supports_response_schema", return_value=False):
+    with patch("crewai.llm.core.supports_response_schema", return_value=False):
         llm = LLM(model="gemini/gemini-1.5-pro", response_format=DummyResponse, is_litellm=True)
         with pytest.raises(ValueError) as excinfo:
             llm._validate_call_params()
@@ -342,7 +342,7 @@ def test_context_window_validation():
     # Test invalid window size
     with pytest.raises(ValueError) as excinfo:
         with patch.dict(
-            "crewai.llm.LLM_CONTEXT_WINDOW_SIZES",
+            "crewai.llm.core.LLM_CONTEXT_WINDOW_SIZES",
             {"test-model": 500},  # Below minimum
             clear=True,
         ):
@@ -702,8 +702,8 @@ def test_ollama_does_not_modify_when_last_is_user(ollama_llm):
 
 def test_native_provider_raises_error_when_supported_but_fails():
     """Test that when a native provider is in SUPPORTED_NATIVE_PROVIDERS but fails to instantiate, we raise the error."""
-    with patch("crewai.llm.SUPPORTED_NATIVE_PROVIDERS", ["openai"]):
-        with patch("crewai.llm.LLM._get_native_provider") as mock_get_native:
+    with patch("crewai.llm.internal.meta.SUPPORTED_NATIVE_PROVIDERS", ["openai"]):
+        with patch("crewai.llm.internal.meta.LLMMeta._get_native_provider") as mock_get_native:
             # Mock that provider exists but throws an error when instantiated
             mock_provider = MagicMock()
             mock_provider.side_effect = ValueError("Native provider initialization failed")
@@ -718,7 +718,7 @@ def test_native_provider_raises_error_when_supported_but_fails():
 
 def test_native_provider_falls_back_to_litellm_when_not_in_supported_list():
     """Test that when a provider is not in SUPPORTED_NATIVE_PROVIDERS, we fall back to LiteLLM."""
-    with patch("crewai.llm.SUPPORTED_NATIVE_PROVIDERS", ["openai", "anthropic"]):
+    with patch("crewai.llm.internal.meta.SUPPORTED_NATIVE_PROVIDERS", ["openai", "anthropic"]):
         # Using a provider not in the supported list
         llm = LLM(model="groq/llama-3.1-70b-versatile", is_litellm=False)
 

lib/crewai/tests/utilities/test_events.py

@@ -743,7 +743,7 @@ def test_llm_emits_call_failed_event():
     error_message = "OpenAI API call failed: Simulated API failure"
 
     with patch(
-        "crewai.llms.providers.openai.completion.OpenAICompletion._handle_completion"
+        "crewai.llm.providers.openai.completion.OpenAICompletion._handle_completion"
     ) as mock_handle_completion:
         mock_handle_completion.side_effect = Exception("Simulated API failure")
 

lib/crewai/tests/utilities/test_llm_utils.py

@@ -4,7 +4,7 @@ from unittest.mock import patch
 
 from crewai.cli.constants import DEFAULT_LLM_MODEL
 from crewai.llm import LLM
-from crewai.llms.base_llm import BaseLLM
+from crewai.llm.base_llm import BaseLLM
 from crewai.utilities.llm_utils import create_llm
 import pytest