Add support for custom LLM implementations

Co-Authored-By: Joe Moura <joao@crewai.com>

src/crewai/__init__.py

@@ -4,7 +4,7 @@ from crewai.agent import Agent
 from crewai.crew import Crew
 from crewai.flow.flow import Flow
 from crewai.knowledge.knowledge import Knowledge
-from crewai.llm import LLM
+from crewai.llm import BaseLLM, LLM
 from crewai.process import Process
 from crewai.task import Task
 
@@ -21,6 +21,7 @@ __all__ = [
     "Process",
     "Task",
     "LLM",
+    "BaseLLM",
     "Flow",
     "Knowledge",
 ]

src/crewai/agent.py

@@ -11,7 +11,7 @@ from crewai.agents.crew_agent_executor import CrewAgentExecutor
 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.llm import LLM
+from crewai.llm import BaseLLM, LLM
 from crewai.memory.contextual.contextual_memory import ContextualMemory
 from crewai.task import Task
 from crewai.tools import BaseTool
@@ -70,10 +70,10 @@ class Agent(BaseAgent):
         default=True,
         description="Use system prompt for the agent.",
     )
-    llm: Union[str, InstanceOf[LLM], Any] = Field(
+    llm: Union[str, InstanceOf[BaseLLM], Any] = Field(
         description="Language model that will run the agent.", default=None
     )
-    function_calling_llm: Optional[Union[str, InstanceOf[LLM], Any]] = Field(
+    function_calling_llm: Optional[Union[str, InstanceOf[BaseLLM], Any]] = Field(
         description="Language model that will run the agent.", default=None
     )
     system_template: Optional[str] = Field(
@@ -117,7 +117,7 @@ class Agent(BaseAgent):
         self.agent_ops_agent_name = self.role
 
         self.llm = create_llm(self.llm)
-        if self.function_calling_llm and not isinstance(self.function_calling_llm, LLM):
+        if self.function_calling_llm and not isinstance(self.function_calling_llm, BaseLLM):
             self.function_calling_llm = create_llm(self.function_calling_llm)
 
         if not self.agent_executor:

src/crewai/crew.py

@@ -26,7 +26,7 @@ from crewai.agents.cache import CacheHandler
 from crewai.crews.crew_output import CrewOutput
 from crewai.knowledge.knowledge import Knowledge
 from crewai.knowledge.source.base_knowledge_source import BaseKnowledgeSource
-from crewai.llm import LLM
+from crewai.llm import BaseLLM, LLM
 from crewai.memory.entity.entity_memory import EntityMemory
 from crewai.memory.long_term.long_term_memory import LongTermMemory
 from crewai.memory.short_term.short_term_memory import ShortTermMemory
@@ -150,7 +150,7 @@ class Crew(BaseModel):
         default=None,
         description="Metrics for the LLM usage during all tasks execution.",
     )
-    manager_llm: Optional[Any] = Field(
+    manager_llm: Optional[Union[str, InstanceOf[BaseLLM], Any]] = Field(
         description="Language model that will run the agent.", default=None
     )
     manager_agent: Optional[BaseAgent] = Field(
@@ -196,7 +196,7 @@ class Crew(BaseModel):
         default=False,
         description="Plan the crew execution and add the plan to the crew.",
     )
-    planning_llm: Optional[Any] = Field(
+    planning_llm: Optional[Union[str, InstanceOf[BaseLLM], Any]] = Field(
         default=None,
         description="Language model that will run the AgentPlanner if planning is True.",
     )
@@ -212,7 +212,7 @@ class Crew(BaseModel):
         default=None,
         description="Knowledge sources for the crew. Add knowledge sources to the knowledge object.",
     )
-    chat_llm: Optional[Any] = Field(
+    chat_llm: Optional[Union[str, InstanceOf[BaseLLM], Any]] = Field(
         default=None,
         description="LLM used to handle chatting with the crew.",
     )
@@ -1193,7 +1193,7 @@ class Crew(BaseModel):
     def test(
         self,
         n_iterations: int,
-        eval_llm: Union[str, InstanceOf[LLM]],
+        eval_llm: Union[str, InstanceOf[BaseLLM]],
         inputs: Optional[Dict[str, Any]] = None,
     ) -> None:
         """Test and evaluate the Crew with the given inputs for n iterations concurrently using concurrent.futures."""

src/crewai/llm.py

@@ -4,6 +4,7 @@ import os
 import sys
 import threading
 import warnings
+from abc import ABC, abstractmethod
 from contextlib import contextmanager
 from typing import Any, Dict, List, Literal, Optional, Type, Union, cast
 
@@ -34,6 +35,78 @@ from crewai.utilities.exceptions.context_window_exceeding_exception import (
 load_dotenv()
 
 
+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.
+    """
+    
+    def __init__(self):
+        """Initialize the BaseLLM with default attributes.
+        
+        This constructor sets default values for attributes that are expected
+        by the CrewAgentExecutor and other components.
+        """
+        self.stop = []
+    
+    @abstractmethod
+    def call(
+        self,
+        messages: Union[str, List[Dict[str, str]]],
+        tools: Optional[List[dict]] = None,
+        callbacks: Optional[List[Any]] = None,
+        available_functions: Optional[Dict[str, Any]] = None,
+    ) -> Union[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.
+            
+        Returns:
+            Either a text response from the LLM (str) or
+            the result of a tool function call (Any).
+        """
+        pass
+        
+    @abstractmethod
+    def supports_function_calling(self) -> bool:
+        """Check if the LLM supports function calling.
+        
+        Returns:
+            True if the LLM supports function calling, False otherwise.
+        """
+        pass
+        
+    @abstractmethod
+    def supports_stop_words(self) -> bool:
+        """Check if the LLM supports stop words.
+        
+        Returns:
+            True if the LLM supports stop words, False otherwise.
+        """
+        pass
+        
+    @abstractmethod
+    def get_context_window_size(self) -> int:
+        """Get the context window size of the LLM.
+        
+        Returns:
+            The context window size as an integer.
+        """
+        pass
+
+
 class FilteredStream:
     def __init__(self, original_stream):
         self._original_stream = original_stream
@@ -126,7 +199,7 @@ def suppress_warnings():
             sys.stderr = old_stderr
 
 
-class LLM:
+class LLM(BaseLLM):
     def __init__(
         self,
         model: str,

src/crewai/utilities/llm_utils.py

@@ -2,28 +2,28 @@ import os
 from typing import Any, Dict, List, Optional, Union
 
 from crewai.cli.constants import DEFAULT_LLM_MODEL, ENV_VARS, LITELLM_PARAMS
-from crewai.llm import LLM
+from crewai.llm import BaseLLM, LLM
 
 
 def create_llm(
-    llm_value: Union[str, LLM, Any, None] = None,
-) -> Optional[LLM]:
+    llm_value: Union[str, BaseLLM, Any, None] = None,
+) -> Optional[BaseLLM]:
     """
     Creates or returns an LLM instance based on the given llm_value.
 
     Args:
-        llm_value (str | LLM | Any | None):
+        llm_value (str | BaseLLM | Any | None):
             - str: The model name (e.g., "gpt-4").
-            - LLM: Already instantiated LLM, returned as-is.
+            - BaseLLM: Already instantiated BaseLLM (including LLM), returned as-is.
             - Any: Attempt to extract known attributes like model_name, temperature, etc.
             - None: Use environment-based or fallback default model.
 
     Returns:
-        An LLM instance if successful, or None if something fails.
+        A BaseLLM instance if successful, or None if something fails.
     """
 
-    # 1) If llm_value is already an LLM object, return it directly
-    if isinstance(llm_value, LLM):
+    # 1) If llm_value is already a BaseLLM object, return it directly
+    if isinstance(llm_value, BaseLLM):
         return llm_value
 
     # 2) If llm_value is a string (model name)