fix: relax openai version constraint to allow newer SDK versions

- Change openai dependency from ~=1.83.0 to >=1.83.0,<2
- Add tests to verify OpenAI SDK imports compatibility
- Add test to verify OpenAI client instantiation

Fixes #4300

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

lib/crewai-files/src/crewai_files/__init__.py

@@ -152,4 +152,4 @@ __all__ = [
     "wrap_file_source",
 ]
 
-__version__ = "1.9.1"
+__version__ = "1.9.2"

lib/crewai-tools/pyproject.toml

@@ -12,7 +12,7 @@ dependencies = [
     "pytube~=15.0.0",
     "requests~=2.32.5",
     "docker~=7.1.0",
-    "crewai==1.9.1",
+    "crewai==1.9.2",
     "lancedb~=0.5.4",
     "tiktoken~=0.8.0",
     "beautifulsoup4~=4.13.4",

lib/crewai-tools/src/crewai_tools/__init__.py

@@ -291,4 +291,4 @@ __all__ = [
     "ZapierActionTools",
 ]
 
-__version__ = "1.9.1"
+__version__ = "1.9.2"

lib/crewai/pyproject.toml

@@ -10,7 +10,7 @@ requires-python = ">=3.10, <3.14"
 dependencies = [
     # Core Dependencies
     "pydantic~=2.11.9",
-    "openai~=1.83.0",
+    "openai>=1.83.0,<2",
     "instructor>=1.3.3",
     # Text Processing
     "pdfplumber~=0.11.4",
@@ -49,7 +49,7 @@ Repository = "https://github.com/crewAIInc/crewAI"
 
 [project.optional-dependencies]
 tools = [
-    "crewai-tools==1.9.1",
+    "crewai-tools==1.9.2",
 ]
 embeddings = [
     "tiktoken~=0.8.0"

lib/crewai/src/crewai/__init__.py

@@ -40,7 +40,7 @@ def _suppress_pydantic_deprecation_warnings() -> None:
 
 _suppress_pydantic_deprecation_warnings()
 
-__version__ = "1.9.1"
+__version__ = "1.9.2"
 _telemetry_submitted = False
 
 

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

@@ -37,7 +37,8 @@ class CrewAgentExecutorMixin:
             self.crew
             and self.agent
             and self.task
-            and f"Action: {sanitize_tool_name('Delegate work to coworker')}" not in output.text
+            and f"Action: {sanitize_tool_name('Delegate work to coworker')}"
+            not in output.text
         ):
             try:
                 if (
@@ -132,10 +133,11 @@ class CrewAgentExecutorMixin:
             and self.crew._long_term_memory
             and self.crew._entity_memory is None
         ):
-            self._printer.print(
-                content="Long term memory is enabled, but entity memory is not enabled. Please configure entity memory or set memory=True to automatically enable it.",
-                color="bold_yellow",
-            )
+            if self.agent and self.agent.verbose:
+                self._printer.print(
+                    content="Long term memory is enabled, but entity memory is not enabled. Please configure entity memory or set memory=True to automatically enable it.",
+                    color="bold_yellow",
+                )
 
     def _ask_human_input(self, final_answer: str) -> str:
         """Prompt human input with mode-appropriate messaging.

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

@@ -206,13 +206,14 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
         try:
             formatted_answer = self._invoke_loop()
         except AssertionError:
-            self._printer.print(
-                content="Agent failed to reach a final answer. This is likely a bug - please report it.",
-                color="red",
-            )
+            if self.agent.verbose:
+                self._printer.print(
+                    content="Agent failed to reach a final answer. This is likely a bug - please report it.",
+                    color="red",
+                )
             raise
         except Exception as e:
-            handle_unknown_error(self._printer, e)
+            handle_unknown_error(self._printer, e, verbose=self.agent.verbose)
             raise
 
         if self.ask_for_human_input:
@@ -327,6 +328,7 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
                         messages=self.messages,
                         llm=self.llm,
                         callbacks=self.callbacks,
+                        verbose=self.agent.verbose,
                     )
                     break
 
@@ -341,22 +343,41 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
                     from_agent=self.agent,
                     response_model=self.response_model,
                     executor_context=self,
+                    verbose=self.agent.verbose,
                 )
                 # breakpoint()
                 if self.response_model is not None:
                     try:
-                        self.response_model.model_validate_json(answer)
-                        formatted_answer = AgentFinish(
-                            thought="",
-                            output=answer,
-                            text=answer,
-                        )
+                        if isinstance(answer, BaseModel):
+                            output_json = answer.model_dump_json()
+                            formatted_answer = AgentFinish(
+                                thought="",
+                                output=answer,
+                                text=output_json,
+                            )
+                        else:
+                            self.response_model.model_validate_json(answer)
+                            formatted_answer = AgentFinish(
+                                thought="",
+                                output=answer,
+                                text=answer,
+                            )
                     except ValidationError:
+                        # If validation fails, convert BaseModel to JSON string for parsing
+                        answer_str = (
+                            answer.model_dump_json()
+                            if isinstance(answer, BaseModel)
+                            else str(answer)
+                        )
                         formatted_answer = process_llm_response(
-                            answer, self.use_stop_words
+                            answer_str, self.use_stop_words
                         )  # type: ignore[assignment]
                 else:
-                    formatted_answer = process_llm_response(answer, self.use_stop_words)  # type: ignore[assignment]
+                    # When no response_model, answer should be a string
+                    answer_str = str(answer) if not isinstance(answer, str) else answer
+                    formatted_answer = process_llm_response(
+                        answer_str, self.use_stop_words
+                    )  # type: ignore[assignment]
 
                 if isinstance(formatted_answer, AgentAction):
                     # Extract agent fingerprint if available
@@ -399,6 +420,7 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
                     iterations=self.iterations,
                     log_error_after=self.log_error_after,
                     printer=self._printer,
+                    verbose=self.agent.verbose,
                 )
 
             except Exception as e:
@@ -413,9 +435,10 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
                         llm=self.llm,
                         callbacks=self.callbacks,
                         i18n=self._i18n,
+                        verbose=self.agent.verbose,
                     )
                     continue
-                handle_unknown_error(self._printer, e)
+                handle_unknown_error(self._printer, e, verbose=self.agent.verbose)
                 raise e
             finally:
                 self.iterations += 1
@@ -461,6 +484,7 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
                         messages=self.messages,
                         llm=self.llm,
                         callbacks=self.callbacks,
+                        verbose=self.agent.verbose,
                     )
                     self._show_logs(formatted_answer)
                     return formatted_answer
@@ -482,6 +506,7 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
                     from_agent=self.agent,
                     response_model=self.response_model,
                     executor_context=self,
+                    verbose=self.agent.verbose,
                 )
 
                 # Check if the response is a list of tool calls
@@ -513,6 +538,18 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
                     self._show_logs(formatted_answer)
                     return formatted_answer
 
+                if isinstance(answer, BaseModel):
+                    output_json = answer.model_dump_json()
+                    formatted_answer = AgentFinish(
+                        thought="",
+                        output=answer,
+                        text=output_json,
+                    )
+                    self._invoke_step_callback(formatted_answer)
+                    self._append_message(output_json)
+                    self._show_logs(formatted_answer)
+                    return formatted_answer
+
                 # Unexpected response type, treat as final answer
                 formatted_answer = AgentFinish(
                     thought="",
@@ -535,9 +572,10 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
                         llm=self.llm,
                         callbacks=self.callbacks,
                         i18n=self._i18n,
+                        verbose=self.agent.verbose,
                     )
                     continue
-                handle_unknown_error(self._printer, e)
+                handle_unknown_error(self._printer, e, verbose=self.agent.verbose)
                 raise e
             finally:
                 self.iterations += 1
@@ -559,13 +597,23 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
             from_agent=self.agent,
             response_model=self.response_model,
             executor_context=self,
+            verbose=self.agent.verbose,
         )
 
-        formatted_answer = AgentFinish(
-            thought="",
-            output=str(answer),
-            text=str(answer),
-        )
+        if isinstance(answer, BaseModel):
+            output_json = answer.model_dump_json()
+            formatted_answer = AgentFinish(
+                thought="",
+                output=answer,
+                text=output_json,
+            )
+        else:
+            answer_str = answer if isinstance(answer, str) else str(answer)
+            formatted_answer = AgentFinish(
+                thought="",
+                output=answer_str,
+                text=answer_str,
+            )
         self._show_logs(formatted_answer)
         return formatted_answer
 
@@ -755,10 +803,10 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
         track_delegation_if_needed(func_name, args_dict, self.task)
 
         # Find the structured tool for hook context
-        structured_tool = None
-        for tool in self.tools or []:
-            if sanitize_tool_name(tool.name) == func_name:
-                structured_tool = tool
+        structured_tool: CrewStructuredTool | None = None
+        for structured in self.tools or []:
+            if sanitize_tool_name(structured.name) == func_name:
+                structured_tool = structured
                 break
 
         # Execute before_tool_call hooks
@@ -779,10 +827,11 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
                     hook_blocked = True
                     break
         except Exception as hook_error:
-            self._printer.print(
-                content=f"Error in before_tool_call hook: {hook_error}",
-                color="red",
-            )
+            if self.agent.verbose:
+                self._printer.print(
+                    content=f"Error in before_tool_call hook: {hook_error}",
+                    color="red",
+                )
 
         # If hook blocked execution, set result and skip tool execution
         if hook_blocked:
@@ -848,15 +897,16 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
         after_hooks = get_after_tool_call_hooks()
         try:
             for after_hook in after_hooks:
-                hook_result = after_hook(after_hook_context)
-                if hook_result is not None:
-                    result = hook_result
+                after_hook_result = after_hook(after_hook_context)
+                if after_hook_result is not None:
+                    result = after_hook_result
                     after_hook_context.tool_result = result
         except Exception as hook_error:
-            self._printer.print(
-                content=f"Error in after_tool_call hook: {hook_error}",
-                color="red",
-            )
+            if self.agent.verbose:
+                self._printer.print(
+                    content=f"Error in after_tool_call hook: {hook_error}",
+                    color="red",
+                )
 
         # Emit tool usage finished event
         crewai_event_bus.emit(
@@ -942,13 +992,14 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
         try:
             formatted_answer = await self._ainvoke_loop()
         except AssertionError:
-            self._printer.print(
-                content="Agent failed to reach a final answer. This is likely a bug - please report it.",
-                color="red",
-            )
+            if self.agent.verbose:
+                self._printer.print(
+                    content="Agent failed to reach a final answer. This is likely a bug - please report it.",
+                    color="red",
+                )
             raise
         except Exception as e:
-            handle_unknown_error(self._printer, e)
+            handle_unknown_error(self._printer, e, verbose=self.agent.verbose)
             raise
 
         if self.ask_for_human_input:
@@ -999,6 +1050,7 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
                         messages=self.messages,
                         llm=self.llm,
                         callbacks=self.callbacks,
+                        verbose=self.agent.verbose,
                     )
                     break
 
@@ -1013,22 +1065,41 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
                     from_agent=self.agent,
                     response_model=self.response_model,
                     executor_context=self,
+                    verbose=self.agent.verbose,
                 )
 
                 if self.response_model is not None:
                     try:
-                        self.response_model.model_validate_json(answer)
-                        formatted_answer = AgentFinish(
-                            thought="",
-                            output=answer,
-                            text=answer,
-                        )
+                        if isinstance(answer, BaseModel):
+                            output_json = answer.model_dump_json()
+                            formatted_answer = AgentFinish(
+                                thought="",
+                                output=answer,
+                                text=output_json,
+                            )
+                        else:
+                            self.response_model.model_validate_json(answer)
+                            formatted_answer = AgentFinish(
+                                thought="",
+                                output=answer,
+                                text=answer,
+                            )
                     except ValidationError:
+                        # If validation fails, convert BaseModel to JSON string for parsing
+                        answer_str = (
+                            answer.model_dump_json()
+                            if isinstance(answer, BaseModel)
+                            else str(answer)
+                        )
                         formatted_answer = process_llm_response(
-                            answer, self.use_stop_words
+                            answer_str, self.use_stop_words
                         )  # type: ignore[assignment]
                 else:
-                    formatted_answer = process_llm_response(answer, self.use_stop_words)  # type: ignore[assignment]
+                    # When no response_model, answer should be a string
+                    answer_str = str(answer) if not isinstance(answer, str) else answer
+                    formatted_answer = process_llm_response(
+                        answer_str, self.use_stop_words
+                    )  # type: ignore[assignment]
 
                 if isinstance(formatted_answer, AgentAction):
                     fingerprint_context = {}
@@ -1070,6 +1141,7 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
                     iterations=self.iterations,
                     log_error_after=self.log_error_after,
                     printer=self._printer,
+                    verbose=self.agent.verbose,
                 )
 
             except Exception as e:
@@ -1083,9 +1155,10 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
                         llm=self.llm,
                         callbacks=self.callbacks,
                         i18n=self._i18n,
+                        verbose=self.agent.verbose,
                     )
                     continue
-                handle_unknown_error(self._printer, e)
+                handle_unknown_error(self._printer, e, verbose=self.agent.verbose)
                 raise e
             finally:
                 self.iterations += 1
@@ -1125,6 +1198,7 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
                         messages=self.messages,
                         llm=self.llm,
                         callbacks=self.callbacks,
+                        verbose=self.agent.verbose,
                     )
                     self._show_logs(formatted_answer)
                     return formatted_answer
@@ -1146,6 +1220,7 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
                     from_agent=self.agent,
                     response_model=self.response_model,
                     executor_context=self,
+                    verbose=self.agent.verbose,
                 )
                 # Check if the response is a list of tool calls
                 if (
@@ -1176,6 +1251,18 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
                     self._show_logs(formatted_answer)
                     return formatted_answer
 
+                if isinstance(answer, BaseModel):
+                    output_json = answer.model_dump_json()
+                    formatted_answer = AgentFinish(
+                        thought="",
+                        output=answer,
+                        text=output_json,
+                    )
+                    self._invoke_step_callback(formatted_answer)
+                    self._append_message(output_json)
+                    self._show_logs(formatted_answer)
+                    return formatted_answer
+
                 # Unexpected response type, treat as final answer
                 formatted_answer = AgentFinish(
                     thought="",
@@ -1198,9 +1285,10 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
                         llm=self.llm,
                         callbacks=self.callbacks,
                         i18n=self._i18n,
+                        verbose=self.agent.verbose,
                     )
                     continue
-                handle_unknown_error(self._printer, e)
+                handle_unknown_error(self._printer, e, verbose=self.agent.verbose)
                 raise e
             finally:
                 self.iterations += 1
@@ -1222,13 +1310,23 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
             from_agent=self.agent,
             response_model=self.response_model,
             executor_context=self,
+            verbose=self.agent.verbose,
         )
 
-        formatted_answer = AgentFinish(
-            thought="",
-            output=str(answer),
-            text=str(answer),
-        )
+        if isinstance(answer, BaseModel):
+            output_json = answer.model_dump_json()
+            formatted_answer = AgentFinish(
+                thought="",
+                output=answer,
+                text=output_json,
+            )
+        else:
+            answer_str = answer if isinstance(answer, str) else str(answer)
+            formatted_answer = AgentFinish(
+                thought="",
+                output=answer_str,
+                text=answer_str,
+            )
         self._show_logs(formatted_answer)
         return formatted_answer
 
@@ -1339,10 +1437,11 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
         )
 
         if train_iteration is None or not isinstance(train_iteration, int):
-            self._printer.print(
-                content="Invalid or missing train iteration. Cannot save training data.",
-                color="red",
-            )
+            if self.agent.verbose:
+                self._printer.print(
+                    content="Invalid or missing train iteration. Cannot save training data.",
+                    color="red",
+                )
             return
 
         training_handler = CrewTrainingHandler(TRAINING_DATA_FILE)
@@ -1362,13 +1461,14 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
             if train_iteration in agent_training_data:
                 agent_training_data[train_iteration]["improved_output"] = result.output
             else:
-                self._printer.print(
-                    content=(
-                        f"No existing training data for agent {agent_id} and iteration "
-                        f"{train_iteration}. Cannot save improved output."
-                    ),
-                    color="red",
-                )
+                if self.agent.verbose:
+                    self._printer.print(
+                        content=(
+                            f"No existing training data for agent {agent_id} and iteration "
+                            f"{train_iteration}. Cannot save improved output."
+                        ),
+                        color="red",
+                    )
                 return
 
         # Update the training data and save
@@ -1399,7 +1499,12 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
         Returns:
             Final answer after feedback.
         """
-        human_feedback = self._ask_human_input(formatted_answer.output)
+        output_str = (
+            formatted_answer.output
+            if isinstance(formatted_answer.output, str)
+            else formatted_answer.output.model_dump_json()
+        )
+        human_feedback = self._ask_human_input(output_str)
 
         if self._is_training_mode():
             return self._handle_training_feedback(formatted_answer, human_feedback)
@@ -1458,7 +1563,12 @@ class CrewAgentExecutor(CrewAgentExecutorMixin):
                 self.ask_for_human_input = False
             else:
                 answer = self._process_feedback_iteration(feedback)
-                feedback = self._ask_human_input(answer.output)
+                output_str = (
+                    answer.output
+                    if isinstance(answer.output, str)
+                    else answer.output.model_dump_json()
+                )
+                feedback = self._ask_human_input(output_str)
 
         return answer
 

lib/crewai/src/crewai/agents/parser.py

@@ -8,6 +8,7 @@ AgentAction or AgentFinish objects.
 from dataclasses import dataclass
 
 from json_repair import repair_json  # type: ignore[import-untyped]
+from pydantic import BaseModel
 
 from crewai.agents.constants import (
     ACTION_INPUT_ONLY_REGEX,
@@ -40,7 +41,7 @@ class AgentFinish:
     """Represents the final answer from an agent."""
 
     thought: str
-    output: str
+    output: str | BaseModel
     text: str
 
 

lib/crewai/src/crewai/cli/templates/crew/pyproject.toml

@@ -5,7 +5,7 @@ description = "{{name}} using crewAI"
 authors = [{ name = "Your Name", email = "you@example.com" }]
 requires-python = ">=3.10,<3.14"
 dependencies = [
-    "crewai[tools]==1.9.1"
+    "crewai[tools]==1.9.2"
 ]
 
 [project.scripts]

lib/crewai/src/crewai/cli/templates/flow/pyproject.toml

@@ -5,7 +5,7 @@ description = "{{name}} using crewAI"
 authors = [{ name = "Your Name", email = "you@example.com" }]
 requires-python = ">=3.10,<3.14"
 dependencies = [
-    "crewai[tools]==1.9.1"
+    "crewai[tools]==1.9.2"
 ]
 
 [project.scripts]

lib/crewai/src/crewai/experimental/agent_executor.py

@@ -341,6 +341,7 @@ class AgentExecutor(Flow[AgentReActState], CrewAgentExecutorMixin):
             messages=list(self.state.messages),
             llm=self.llm,
             callbacks=self.callbacks,
+            verbose=self.agent.verbose,
         )
 
         self.state.current_answer = formatted_answer
@@ -366,6 +367,7 @@ class AgentExecutor(Flow[AgentReActState], CrewAgentExecutorMixin):
                 from_agent=self.agent,
                 response_model=None,
                 executor_context=self,
+                verbose=self.agent.verbose,
             )
 
             # Parse the LLM response
@@ -401,7 +403,7 @@ class AgentExecutor(Flow[AgentReActState], CrewAgentExecutorMixin):
                 return "context_error"
             if e.__class__.__module__.startswith("litellm"):
                 raise e
-            handle_unknown_error(self._printer, e)
+            handle_unknown_error(self._printer, e, verbose=self.agent.verbose)
             raise
 
     @listen("continue_reasoning_native")
@@ -436,6 +438,7 @@ class AgentExecutor(Flow[AgentReActState], CrewAgentExecutorMixin):
                 from_agent=self.agent,
                 response_model=None,
                 executor_context=self,
+                verbose=self.agent.verbose,
             )
 
             # Check if the response is a list of tool calls
@@ -474,7 +477,7 @@ class AgentExecutor(Flow[AgentReActState], CrewAgentExecutorMixin):
                 return "context_error"
             if e.__class__.__module__.startswith("litellm"):
                 raise e
-            handle_unknown_error(self._printer, e)
+            handle_unknown_error(self._printer, e, verbose=self.agent.verbose)
             raise
 
     @router(call_llm_and_parse)
@@ -670,10 +673,10 @@ class AgentExecutor(Flow[AgentReActState], CrewAgentExecutorMixin):
 
             track_delegation_if_needed(func_name, args_dict, self.task)
 
-            structured_tool = None
-            for tool in self.tools or []:
-                if sanitize_tool_name(tool.name) == func_name:
-                    structured_tool = tool
+            structured_tool: CrewStructuredTool | None = None
+            for structured in self.tools or []:
+                if sanitize_tool_name(structured.name) == func_name:
+                    structured_tool = structured
                     break
 
             hook_blocked = False
@@ -693,10 +696,11 @@ class AgentExecutor(Flow[AgentReActState], CrewAgentExecutorMixin):
                         hook_blocked = True
                         break
             except Exception as hook_error:
-                self._printer.print(
-                    content=f"Error in before_tool_call hook: {hook_error}",
-                    color="red",
-                )
+                if self.agent.verbose:
+                    self._printer.print(
+                        content=f"Error in before_tool_call hook: {hook_error}",
+                        color="red",
+                    )
 
             if hook_blocked:
                 result = f"Tool execution blocked by hook. Tool: {func_name}"
@@ -758,15 +762,16 @@ class AgentExecutor(Flow[AgentReActState], CrewAgentExecutorMixin):
             after_hooks = get_after_tool_call_hooks()
             try:
                 for after_hook in after_hooks:
-                    hook_result = after_hook(after_hook_context)
-                    if hook_result is not None:
-                        result = hook_result
+                    after_hook_result = after_hook(after_hook_context)
+                    if after_hook_result is not None:
+                        result = after_hook_result
                         after_hook_context.tool_result = result
             except Exception as hook_error:
-                self._printer.print(
-                    content=f"Error in after_tool_call hook: {hook_error}",
-                    color="red",
-                )
+                if self.agent.verbose:
+                    self._printer.print(
+                        content=f"Error in after_tool_call hook: {hook_error}",
+                        color="red",
+                    )
 
             # Emit tool usage finished event
             crewai_event_bus.emit(
@@ -814,15 +819,6 @@ class AgentExecutor(Flow[AgentReActState], CrewAgentExecutorMixin):
                 self.state.is_finished = True
                 return "tool_result_is_final"
 
-        # Add reflection prompt once after all tools in the batch
-        reasoning_prompt = self._i18n.slice("post_tool_reasoning")
-
-        reasoning_message: LLMMessage = {
-            "role": "user",
-            "content": reasoning_prompt,
-        }
-        self.state.messages.append(reasoning_message)
-
         return "native_tool_completed"
 
     def _extract_tool_name(self, tool_call: Any) -> str:
@@ -911,6 +907,7 @@ class AgentExecutor(Flow[AgentReActState], CrewAgentExecutorMixin):
             iterations=self.state.iterations,
             log_error_after=self.log_error_after,
             printer=self._printer,
+            verbose=self.agent.verbose,
         )
 
         if formatted_answer:
@@ -930,6 +927,7 @@ class AgentExecutor(Flow[AgentReActState], CrewAgentExecutorMixin):
             llm=self.llm,
             callbacks=self.callbacks,
             i18n=self._i18n,
+            verbose=self.agent.verbose,
         )
 
         self.state.iterations += 1
@@ -1021,7 +1019,7 @@ class AgentExecutor(Flow[AgentReActState], CrewAgentExecutorMixin):
             self._console.print(fail_text)
             raise
         except Exception as e:
-            handle_unknown_error(self._printer, e)
+            handle_unknown_error(self._printer, e, verbose=self.agent.verbose)
             raise
         finally:
             self._is_executing = False
@@ -1106,7 +1104,7 @@ class AgentExecutor(Flow[AgentReActState], CrewAgentExecutorMixin):
             self._console.print(fail_text)
             raise
         except Exception as e:
-            handle_unknown_error(self._printer, e)
+            handle_unknown_error(self._printer, e, verbose=self.agent.verbose)
             raise
         finally:
             self._is_executing = False

lib/crewai/src/crewai/flow/persistence/decorators.py

@@ -118,17 +118,20 @@ class PersistenceDecorator:
                 )
             except Exception as e:
                 error_msg = LOG_MESSAGES["save_error"].format(method_name, str(e))
-                cls._printer.print(error_msg, color="red")
+                if verbose:
+                    cls._printer.print(error_msg, color="red")
                 logger.error(error_msg)
                 raise RuntimeError(f"State persistence failed: {e!s}") from e
         except AttributeError as e:
             error_msg = LOG_MESSAGES["state_missing"]
-            cls._printer.print(error_msg, color="red")
+            if verbose:
+                cls._printer.print(error_msg, color="red")
             logger.error(error_msg)
             raise ValueError(error_msg) from e
         except (TypeError, ValueError) as e:
             error_msg = LOG_MESSAGES["id_missing"]
-            cls._printer.print(error_msg, color="red")
+            if verbose:
+                cls._printer.print(error_msg, color="red")
             logger.error(error_msg)
             raise ValueError(error_msg) from e
 

lib/crewai/src/crewai/flow/utils.py

@@ -151,7 +151,9 @@ def _unwrap_function(function: Any) -> Any:
     return function
 
 
-def get_possible_return_constants(function: Any) -> list[str] | None:
+def get_possible_return_constants(
+    function: Any, verbose: bool = True
+) -> list[str] | None:
     """Extract possible string return values from a function using AST parsing.
 
     This function analyzes the source code of a router method to identify
@@ -178,10 +180,11 @@ def get_possible_return_constants(function: Any) -> list[str] | None:
         # Can't get source code
         return None
     except Exception as e:
-        _printer.print(
-            f"Error retrieving source code for function {function.__name__}: {e}",
-            color="red",
-        )
+        if verbose:
+            _printer.print(
+                f"Error retrieving source code for function {function.__name__}: {e}",
+                color="red",
+            )
         return None
 
     try:
@@ -190,25 +193,28 @@ def get_possible_return_constants(function: Any) -> list[str] | None:
         # Parse the source code into an AST
         code_ast = ast.parse(source)
     except IndentationError as e:
-        _printer.print(
-            f"IndentationError while parsing source code of {function.__name__}: {e}",
-            color="red",
-        )
-        _printer.print(f"Source code:\n{source}", color="yellow")
+        if verbose:
+            _printer.print(
+                f"IndentationError while parsing source code of {function.__name__}: {e}",
+                color="red",
+            )
+            _printer.print(f"Source code:\n{source}", color="yellow")
         return None
     except SyntaxError as e:
-        _printer.print(
-            f"SyntaxError while parsing source code of {function.__name__}: {e}",
-            color="red",
-        )
-        _printer.print(f"Source code:\n{source}", color="yellow")
+        if verbose:
+            _printer.print(
+                f"SyntaxError while parsing source code of {function.__name__}: {e}",
+                color="red",
+            )
+            _printer.print(f"Source code:\n{source}", color="yellow")
         return None
     except Exception as e:
-        _printer.print(
-            f"Unexpected error while parsing source code of {function.__name__}: {e}",
-            color="red",
-        )
-        _printer.print(f"Source code:\n{source}", color="yellow")
+        if verbose:
+            _printer.print(
+                f"Unexpected error while parsing source code of {function.__name__}: {e}",
+                color="red",
+            )
+            _printer.print(f"Source code:\n{source}", color="yellow")
         return None
 
     return_values: set[str] = set()
@@ -388,15 +394,17 @@ def get_possible_return_constants(function: Any) -> list[str] | None:
 
                 StateAttributeVisitor().visit(class_ast)
             except Exception as e:
-                _printer.print(
-                    f"Could not analyze class context for {function.__name__}: {e}",
-                    color="yellow",
-                )
+                if verbose:
+                    _printer.print(
+                        f"Could not analyze class context for {function.__name__}: {e}",
+                        color="yellow",
+                    )
     except Exception as e:
-        _printer.print(
-            f"Could not introspect class for {function.__name__}: {e}",
-            color="yellow",
-        )
+        if verbose:
+            _printer.print(
+                f"Could not introspect class for {function.__name__}: {e}",
+                color="yellow",
+            )
 
     VariableAssignmentVisitor().visit(code_ast)
     ReturnVisitor().visit(code_ast)

lib/crewai/src/crewai/lite_agent.py

@@ -72,13 +72,13 @@ from crewai.utilities.agent_utils import (
 from crewai.utilities.converter import (
     Converter,
     ConverterError,
-    generate_model_description,
 )
 from crewai.utilities.guardrail import process_guardrail
 from crewai.utilities.guardrail_types import GuardrailCallable, GuardrailType
 from crewai.utilities.i18n import I18N, get_i18n
 from crewai.utilities.llm_utils import create_llm
 from crewai.utilities.printer import Printer
+from crewai.utilities.pydantic_schema_utils import generate_model_description
 from crewai.utilities.token_counter_callback import TokenCalcHandler
 from crewai.utilities.tool_utils import execute_tool_and_check_finality
 from crewai.utilities.types import LLMMessage
@@ -344,11 +344,12 @@ class LiteAgent(FlowTrackable, BaseModel):
             )
 
         except Exception as e:
-            self._printer.print(
-                content="Agent failed to reach a final answer. This is likely a bug - please report it.",
-                color="red",
-            )
-            handle_unknown_error(self._printer, e)
+            if self.verbose:
+                self._printer.print(
+                    content="Agent failed to reach a final answer. This is likely a bug - please report it.",
+                    color="red",
+                )
+            handle_unknown_error(self._printer, e, verbose=self.verbose)
             # Emit error event
             crewai_event_bus.emit(
                 self,
@@ -396,10 +397,11 @@ class LiteAgent(FlowTrackable, BaseModel):
                 if isinstance(result, BaseModel):
                     formatted_result = result
             except ConverterError as e:
-                self._printer.print(
-                    content=f"Failed to parse output into response format after retries: {e.message}",
-                    color="yellow",
-                )
+                if self.verbose:
+                    self._printer.print(
+                        content=f"Failed to parse output into response format after retries: {e.message}",
+                        color="yellow",
+                    )
 
         # Calculate token usage metrics
         if isinstance(self.llm, BaseLLM):
@@ -605,6 +607,7 @@ class LiteAgent(FlowTrackable, BaseModel):
                         messages=self._messages,
                         llm=cast(LLM, self.llm),
                         callbacks=self._callbacks,
+                        verbose=self.verbose,
                     )
 
                 enforce_rpm_limit(self.request_within_rpm_limit)
@@ -617,6 +620,7 @@ class LiteAgent(FlowTrackable, BaseModel):
                         printer=self._printer,
                         from_agent=self,
                         executor_context=self,
+                        verbose=self.verbose,
                     )
 
                 except Exception as e:
@@ -646,16 +650,18 @@ class LiteAgent(FlowTrackable, BaseModel):
 
                 self._append_message(formatted_answer.text, role="assistant")
             except OutputParserError as e:  # noqa: PERF203
-                self._printer.print(
-                    content="Failed to parse LLM output. Retrying...",
-                    color="yellow",
-                )
+                if self.verbose:
+                    self._printer.print(
+                        content="Failed to parse LLM output. Retrying...",
+                        color="yellow",
+                    )
                 formatted_answer = handle_output_parser_exception(
                     e=e,
                     messages=self._messages,
                     iterations=self._iterations,
                     log_error_after=3,
                     printer=self._printer,
+                    verbose=self.verbose,
                 )
 
             except Exception as e:
@@ -670,9 +676,10 @@ class LiteAgent(FlowTrackable, BaseModel):
                         llm=cast(LLM, self.llm),
                         callbacks=self._callbacks,
                         i18n=self.i18n,
+                        verbose=self.verbose,
                     )
                     continue
-                handle_unknown_error(self._printer, e)
+                handle_unknown_error(self._printer, e, verbose=self.verbose)
                 raise e
 
             finally:

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

@@ -404,7 +404,7 @@ class BaseLLM(ABC):
         from_agent: Agent | None = None,
         tool_call: dict[str, Any] | None = None,
         call_type: LLMCallType | None = None,
-        response_id: str | None = None
+        response_id: str | None = None,
     ) -> None:
         """Emit stream chunk event.
 
@@ -427,7 +427,7 @@ class BaseLLM(ABC):
                 from_task=from_task,
                 from_agent=from_agent,
                 call_type=call_type,
-                response_id=response_id
+                response_id=response_id,
             ),
         )
 
@@ -497,7 +497,7 @@ class BaseLLM(ABC):
                 from_agent=from_agent,
             )
 
-            return result
+            return str(result) if not isinstance(result, str) else result
 
         except Exception as e:
             error_msg = f"Error executing function '{function_name}': {e!s}"
@@ -737,22 +737,25 @@ class BaseLLM(ABC):
             task=None,
             crew=None,
         )
+        verbose = getattr(from_agent, "verbose", True) if from_agent else True
         printer = Printer()
 
         try:
             for hook in before_hooks:
                 result = hook(hook_context)
                 if result is False:
-                    printer.print(
-                        content="LLM call blocked by before_llm_call hook",
-                        color="yellow",
-                    )
+                    if verbose:
+                        printer.print(
+                            content="LLM call blocked by before_llm_call hook",
+                            color="yellow",
+                        )
                     return False
         except Exception as e:
-            printer.print(
-                content=f"Error in before_llm_call hook: {e}",
-                color="yellow",
-            )
+            if verbose:
+                printer.print(
+                    content=f"Error in before_llm_call hook: {e}",
+                    color="yellow",
+                )
 
         return True
 
@@ -805,6 +808,7 @@ class BaseLLM(ABC):
             crew=None,
             response=response,
         )
+        verbose = getattr(from_agent, "verbose", True) if from_agent else True
         printer = Printer()
         modified_response = response
 
@@ -815,9 +819,10 @@ class BaseLLM(ABC):
                     modified_response = result
                     hook_context.response = modified_response
         except Exception as e:
-            printer.print(
-                content=f"Error in after_llm_call hook: {e}",
-                color="yellow",
-            )
+            if verbose:
+                printer.print(
+                    content=f"Error in after_llm_call hook: {e}",
+                    color="yellow",
+                )
 
         return modified_response

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

@@ -23,7 +23,7 @@ if TYPE_CHECKING:
 try:
     from anthropic import Anthropic, AsyncAnthropic, transform_schema
     from anthropic.types import Message, TextBlock, ThinkingBlock, ToolUseBlock
-    from anthropic.types.beta import BetaMessage
+    from anthropic.types.beta import BetaMessage, BetaTextBlock
     import httpx
 except ImportError:
     raise ImportError(
@@ -337,6 +337,7 @@ class AnthropicCompletion(BaseLLM):
             available_functions: Available functions for tool calling
             from_task: Task that initiated the call
             from_agent: Agent that initiated the call
+            response_model: Optional response model.
 
         Returns:
             Chat completion response or tool call result
@@ -677,31 +678,31 @@ class AnthropicCompletion(BaseLLM):
         if _is_pydantic_model_class(response_model) and response.content:
             if use_native_structured_output:
                 for block in response.content:
-                    if isinstance(block, TextBlock):
-                        structured_json = block.text
+                    if isinstance(block, (TextBlock, BetaTextBlock)):
+                        structured_data = response_model.model_validate_json(block.text)
                         self._emit_call_completed_event(
-                            response=structured_json,
+                            response=structured_data.model_dump_json(),
                             call_type=LLMCallType.LLM_CALL,
                             from_task=from_task,
                             from_agent=from_agent,
                             messages=params["messages"],
                         )
-                        return structured_json
+                        return structured_data
             else:
                 for block in response.content:
                     if (
                         isinstance(block, ToolUseBlock)
                         and block.name == "structured_output"
                     ):
-                        structured_json = json.dumps(block.input)
+                        structured_data = response_model.model_validate(block.input)
                         self._emit_call_completed_event(
-                            response=structured_json,
+                            response=structured_data.model_dump_json(),
                             call_type=LLMCallType.LLM_CALL,
                             from_task=from_task,
                             from_agent=from_agent,
                             messages=params["messages"],
                         )
-                        return structured_json
+                        return structured_data
 
         # Check if Claude wants to use tools
         if response.content:
@@ -897,28 +898,29 @@ class AnthropicCompletion(BaseLLM):
 
         if _is_pydantic_model_class(response_model):
             if use_native_structured_output:
+                structured_data = response_model.model_validate_json(full_response)
                 self._emit_call_completed_event(
-                    response=full_response,
+                    response=structured_data.model_dump_json(),
                     call_type=LLMCallType.LLM_CALL,
                     from_task=from_task,
                     from_agent=from_agent,
                     messages=params["messages"],
                 )
-                return full_response
+                return structured_data
             for block in final_message.content:
                 if (
                     isinstance(block, ToolUseBlock)
                     and block.name == "structured_output"
                 ):
-                    structured_json = json.dumps(block.input)
+                    structured_data = response_model.model_validate(block.input)
                     self._emit_call_completed_event(
-                        response=structured_json,
+                        response=structured_data.model_dump_json(),
                         call_type=LLMCallType.LLM_CALL,
                         from_task=from_task,
                         from_agent=from_agent,
                         messages=params["messages"],
                     )
-                    return structured_json
+                    return structured_data
 
         if final_message.content:
             tool_uses = [
@@ -1166,31 +1168,31 @@ class AnthropicCompletion(BaseLLM):
         if _is_pydantic_model_class(response_model) and response.content:
             if use_native_structured_output:
                 for block in response.content:
-                    if isinstance(block, TextBlock):
-                        structured_json = block.text
+                    if isinstance(block, (TextBlock, BetaTextBlock)):
+                        structured_data = response_model.model_validate_json(block.text)
                         self._emit_call_completed_event(
-                            response=structured_json,
+                            response=structured_data.model_dump_json(),
                             call_type=LLMCallType.LLM_CALL,
                             from_task=from_task,
                             from_agent=from_agent,
                             messages=params["messages"],
                         )
-                        return structured_json
+                        return structured_data
             else:
                 for block in response.content:
                     if (
                         isinstance(block, ToolUseBlock)
                         and block.name == "structured_output"
                     ):
-                        structured_json = json.dumps(block.input)
+                        structured_data = response_model.model_validate(block.input)
                         self._emit_call_completed_event(
-                            response=structured_json,
+                            response=structured_data.model_dump_json(),
                             call_type=LLMCallType.LLM_CALL,
                             from_task=from_task,
                             from_agent=from_agent,
                             messages=params["messages"],
                         )
-                        return structured_json
+                        return structured_data
 
         if response.content:
             tool_uses = [
@@ -1362,28 +1364,29 @@ class AnthropicCompletion(BaseLLM):
 
         if _is_pydantic_model_class(response_model):
             if use_native_structured_output:
+                structured_data = response_model.model_validate_json(full_response)
                 self._emit_call_completed_event(
-                    response=full_response,
+                    response=structured_data.model_dump_json(),
                     call_type=LLMCallType.LLM_CALL,
                     from_task=from_task,
                     from_agent=from_agent,
                     messages=params["messages"],
                 )
-                return full_response
+                return structured_data
             for block in final_message.content:
                 if (
                     isinstance(block, ToolUseBlock)
                     and block.name == "structured_output"
                 ):
-                    structured_json = json.dumps(block.input)
+                    structured_data = response_model.model_validate(block.input)
                     self._emit_call_completed_event(
-                        response=structured_json,
+                        response=structured_data.model_dump_json(),
                         call_type=LLMCallType.LLM_CALL,
                         from_task=from_task,
                         from_agent=from_agent,
                         messages=params["messages"],
                     )
-                    return structured_json
+                    return structured_data
 
         if final_message.content:
             tool_uses = [

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

@@ -557,7 +557,7 @@ class AzureCompletion(BaseLLM):
         params: AzureCompletionParams,
         from_task: Any | None = None,
         from_agent: Any | None = None,
-    ) -> str:
+    ) -> BaseModel:
         """Validate content against response model and emit completion event.
 
         Args:
@@ -568,24 +568,23 @@ class AzureCompletion(BaseLLM):
             from_agent: Agent that initiated the call
 
         Returns:
-            Validated and serialized JSON string
+            Validated Pydantic model instance
 
         Raises:
             ValueError: If validation fails
         """
         try:
             structured_data = response_model.model_validate_json(content)
-            structured_json = structured_data.model_dump_json()
 
             self._emit_call_completed_event(
-                response=structured_json,
+                response=structured_data.model_dump_json(),
                 call_type=LLMCallType.LLM_CALL,
                 from_task=from_task,
                 from_agent=from_agent,
                 messages=params["messages"],
             )
 
-            return structured_json
+            return structured_data
         except Exception as e:
             error_msg = f"Failed to validate structured output with model {response_model.__name__}: {e}"
             logging.error(error_msg)

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

@@ -132,6 +132,9 @@ class GeminiCompletion(BaseLLM):
         self.supports_tools = bool(
             version_match and float(version_match.group(1)) >= 1.5
         )
+        self.is_gemini_2_0 = bool(
+            version_match and float(version_match.group(1)) >= 2.0
+        )
 
     @property
     def stop(self) -> list[str]:
@@ -439,6 +442,11 @@ class GeminiCompletion(BaseLLM):
 
         Returns:
             GenerateContentConfig object for Gemini API
+
+        Note:
+            Structured output support varies by model version:
+            - Gemini 1.5 and earlier: Uses response_schema (Pydantic model)
+            - Gemini 2.0+: Uses response_json_schema (JSON Schema) with propertyOrdering
         """
         self.tools = tools
         config_params: dict[str, Any] = {}
@@ -466,9 +474,13 @@ class GeminiCompletion(BaseLLM):
         if response_model:
             config_params["response_mime_type"] = "application/json"
             schema_output = generate_model_description(response_model)
-            config_params["response_schema"] = schema_output.get("json_schema", {}).get(
-                "schema", {}
-            )
+            schema = schema_output.get("json_schema", {}).get("schema", {})
+
+            if self.is_gemini_2_0:
+                schema = self._add_property_ordering(schema)
+                config_params["response_json_schema"] = schema
+            else:
+                config_params["response_schema"] = response_model
 
         # Handle tools for supported models
         if tools and self.supports_tools:
@@ -632,7 +644,7 @@ class GeminiCompletion(BaseLLM):
         messages_for_event: list[LLMMessage],
         from_task: Any | None = None,
         from_agent: Any | None = None,
-    ) -> str:
+    ) -> BaseModel:
         """Validate content against response model and emit completion event.
 
         Args:
@@ -643,24 +655,23 @@ class GeminiCompletion(BaseLLM):
             from_agent: Agent that initiated the call
 
         Returns:
-            Validated and serialized JSON string
+            Validated Pydantic model instance
 
         Raises:
             ValueError: If validation fails
         """
         try:
             structured_data = response_model.model_validate_json(content)
-            structured_json = structured_data.model_dump_json()
 
             self._emit_call_completed_event(
-                response=structured_json,
+                response=structured_data.model_dump_json(),
                 call_type=LLMCallType.LLM_CALL,
                 from_task=from_task,
                 from_agent=from_agent,
                 messages=messages_for_event,
             )
 
-            return structured_json
+            return structured_data
         except Exception as e:
             error_msg = f"Failed to validate structured output with model {response_model.__name__}: {e}"
             logging.error(error_msg)
@@ -673,7 +684,7 @@ class GeminiCompletion(BaseLLM):
         response_model: type[BaseModel] | None = None,
         from_task: Any | None = None,
         from_agent: Any | None = None,
-    ) -> str:
+    ) -> str | BaseModel:
         """Finalize completion response with validation and event emission.
 
         Args:
@@ -684,7 +695,7 @@ class GeminiCompletion(BaseLLM):
             from_agent: Agent that initiated the call
 
         Returns:
-            Final response content after processing
+            Final response content after processing (str or Pydantic model if response_model provided)
         """
         messages_for_event = self._convert_contents_to_dict(contents)
 
@@ -870,7 +881,7 @@ class GeminiCompletion(BaseLLM):
         from_task: Any | None = None,
         from_agent: Any | None = None,
         response_model: type[BaseModel] | None = None,
-    ) -> str | list[dict[str, Any]]:
+    ) -> str | BaseModel | list[dict[str, Any]]:
         """Finalize streaming response with usage tracking, function execution, and events.
 
         Args:
@@ -990,7 +1001,7 @@ class GeminiCompletion(BaseLLM):
         from_task: Any | None = None,
         from_agent: Any | None = None,
         response_model: type[BaseModel] | None = None,
-    ) -> str | Any:
+    ) -> str | BaseModel | list[dict[str, Any]] | Any:
         """Handle streaming content generation."""
         full_response = ""
         function_calls: dict[int, dict[str, Any]] = {}
@@ -1190,6 +1201,36 @@ class GeminiCompletion(BaseLLM):
 
         return "".join(text_parts)
 
+    @staticmethod
+    def _add_property_ordering(schema: dict[str, Any]) -> dict[str, Any]:
+        """Add propertyOrdering to JSON schema for Gemini 2.0 compatibility.
+
+        Gemini 2.0 models require an explicit propertyOrdering list to define
+        the preferred structure of JSON objects. This recursively adds
+        propertyOrdering to all objects in the schema.
+
+        Args:
+            schema: JSON schema dictionary.
+
+        Returns:
+            Modified schema with propertyOrdering added to all objects.
+        """
+        if isinstance(schema, dict):
+            if schema.get("type") == "object" and "properties" in schema:
+                properties = schema["properties"]
+                if properties and "propertyOrdering" not in schema:
+                    schema["propertyOrdering"] = list(properties.keys())
+
+            for value in schema.values():
+                if isinstance(value, dict):
+                    GeminiCompletion._add_property_ordering(value)
+                elif isinstance(value, list):
+                    for item in value:
+                        if isinstance(item, dict):
+                            GeminiCompletion._add_property_ordering(item)
+
+        return schema
+
     @staticmethod
     def _convert_contents_to_dict(
         contents: list[types.Content],

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

@@ -1570,15 +1570,14 @@ class OpenAICompletion(BaseLLM):
 
                 parsed_object = parsed_response.choices[0].message.parsed
                 if parsed_object:
-                    structured_json = parsed_object.model_dump_json()
                     self._emit_call_completed_event(
-                        response=structured_json,
+                        response=parsed_object.model_dump_json(),
                         call_type=LLMCallType.LLM_CALL,
                         from_task=from_task,
                         from_agent=from_agent,
                         messages=params["messages"],
                     )
-                    return structured_json
+                    return parsed_object
 
             response: ChatCompletion = self.client.chat.completions.create(**params)
 
@@ -1692,7 +1691,7 @@ class OpenAICompletion(BaseLLM):
         from_task: Any | None = None,
         from_agent: Any | None = None,
         response_model: type[BaseModel] | None = None,
-    ) -> str:
+    ) -> str | BaseModel:
         """Handle streaming chat completion."""
         full_response = ""
         tool_calls: dict[int, dict[str, Any]] = {}
@@ -1728,15 +1727,14 @@ class OpenAICompletion(BaseLLM):
                     if final_completion.choices:
                         parsed_result = final_completion.choices[0].message.parsed
                         if parsed_result:
-                            structured_json = parsed_result.model_dump_json()
                             self._emit_call_completed_event(
-                                response=structured_json,
+                                response=parsed_result.model_dump_json(),
                                 call_type=LLMCallType.LLM_CALL,
                                 from_task=from_task,
                                 from_agent=from_agent,
                                 messages=params["messages"],
                             )
-                            return structured_json
+                            return parsed_result
 
             logging.error("Failed to get parsed result from stream")
             return ""
@@ -1887,15 +1885,14 @@ class OpenAICompletion(BaseLLM):
 
                 parsed_object = parsed_response.choices[0].message.parsed
                 if parsed_object:
-                    structured_json = parsed_object.model_dump_json()
                     self._emit_call_completed_event(
-                        response=structured_json,
+                        response=parsed_object.model_dump_json(),
                         call_type=LLMCallType.LLM_CALL,
                         from_task=from_task,
                         from_agent=from_agent,
                         messages=params["messages"],
                     )
-                    return structured_json
+                    return parsed_object
 
             response: ChatCompletion = await self.async_client.chat.completions.create(
                 **params
@@ -2006,7 +2003,7 @@ class OpenAICompletion(BaseLLM):
         from_task: Any | None = None,
         from_agent: Any | None = None,
         response_model: type[BaseModel] | None = None,
-    ) -> str:
+    ) -> str | BaseModel:
         """Handle async streaming chat completion."""
         full_response = ""
         tool_calls: dict[int, dict[str, Any]] = {}
@@ -2044,17 +2041,16 @@ class OpenAICompletion(BaseLLM):
 
             try:
                 parsed_object = response_model.model_validate_json(accumulated_content)
-                structured_json = parsed_object.model_dump_json()
 
                 self._emit_call_completed_event(
-                    response=structured_json,
+                    response=parsed_object.model_dump_json(),
                     call_type=LLMCallType.LLM_CALL,
                     from_task=from_task,
                     from_agent=from_agent,
                     messages=params["messages"],
                 )
 
-                return structured_json
+                return parsed_object
             except Exception as e:
                 logging.error(f"Failed to parse structured output from stream: {e}")
                 self._emit_call_completed_event(

lib/crewai/src/crewai/memory/storage/ltm_sqlite_storage.py

@@ -12,15 +12,17 @@ from crewai.utilities.paths import db_storage_path
 class LTMSQLiteStorage:
     """SQLite storage class for long-term memory data."""
 
-    def __init__(self, db_path: str | None = None) -> None:
+    def __init__(self, db_path: str | None = None, verbose: bool = True) -> None:
         """Initialize the SQLite storage.
 
         Args:
             db_path: Optional path to the database file.
+            verbose: Whether to print error messages.
         """
         if db_path is None:
             db_path = str(Path(db_storage_path()) / "long_term_memory_storage.db")
         self.db_path = db_path
+        self._verbose = verbose
         self._printer: Printer = Printer()
         Path(self.db_path).parent.mkdir(parents=True, exist_ok=True)
         self._initialize_db()
@@ -44,10 +46,11 @@ class LTMSQLiteStorage:
 
                 conn.commit()
         except sqlite3.Error as e:
-            self._printer.print(
-                content=f"MEMORY ERROR: An error occurred during database initialization: {e}",
-                color="red",
-            )
+            if self._verbose:
+                self._printer.print(
+                    content=f"MEMORY ERROR: An error occurred during database initialization: {e}",
+                    color="red",
+                )
 
     def save(
         self,
@@ -69,10 +72,11 @@ class LTMSQLiteStorage:
                 )
                 conn.commit()
         except sqlite3.Error as e:
-            self._printer.print(
-                content=f"MEMORY ERROR: An error occurred while saving to LTM: {e}",
-                color="red",
-            )
+            if self._verbose:
+                self._printer.print(
+                    content=f"MEMORY ERROR: An error occurred while saving to LTM: {e}",
+                    color="red",
+                )
 
     def load(self, task_description: str, latest_n: int) -> list[dict[str, Any]] | None:
         """Queries the LTM table by task description with error handling."""
@@ -101,10 +105,11 @@ class LTMSQLiteStorage:
                     ]
 
         except sqlite3.Error as e:
-            self._printer.print(
-                content=f"MEMORY ERROR: An error occurred while querying LTM: {e}",
-                color="red",
-            )
+            if self._verbose:
+                self._printer.print(
+                    content=f"MEMORY ERROR: An error occurred while querying LTM: {e}",
+                    color="red",
+                )
         return None
 
     def reset(self) -> None:
@@ -116,10 +121,11 @@ class LTMSQLiteStorage:
                 conn.commit()
 
         except sqlite3.Error as e:
-            self._printer.print(
-                content=f"MEMORY ERROR: An error occurred while deleting all rows in LTM: {e}",
-                color="red",
-            )
+            if self._verbose:
+                self._printer.print(
+                    content=f"MEMORY ERROR: An error occurred while deleting all rows in LTM: {e}",
+                    color="red",
+                )
 
     async def asave(
         self,
@@ -147,10 +153,11 @@ class LTMSQLiteStorage:
                 )
                 await conn.commit()
         except aiosqlite.Error as e:
-            self._printer.print(
-                content=f"MEMORY ERROR: An error occurred while saving to LTM: {e}",
-                color="red",
-            )
+            if self._verbose:
+                self._printer.print(
+                    content=f"MEMORY ERROR: An error occurred while saving to LTM: {e}",
+                    color="red",
+                )
 
     async def aload(
         self, task_description: str, latest_n: int
@@ -187,10 +194,11 @@ class LTMSQLiteStorage:
                         for row in rows
                     ]
         except aiosqlite.Error as e:
-            self._printer.print(
-                content=f"MEMORY ERROR: An error occurred while querying LTM: {e}",
-                color="red",
-            )
+            if self._verbose:
+                self._printer.print(
+                    content=f"MEMORY ERROR: An error occurred while querying LTM: {e}",
+                    color="red",
+                )
         return None
 
     async def areset(self) -> None:
@@ -200,7 +208,8 @@ class LTMSQLiteStorage:
                 await conn.execute("DELETE FROM long_term_memories")
                 await conn.commit()
         except aiosqlite.Error as e:
-            self._printer.print(
-                content=f"MEMORY ERROR: An error occurred while deleting all rows in LTM: {e}",
-                color="red",
-            )
+            if self._verbose:
+                self._printer.print(
+                    content=f"MEMORY ERROR: An error occurred while deleting all rows in LTM: {e}",
+                    color="red",
+                )

lib/crewai/src/crewai/rag/embeddings/providers/ibm/embedding_callable.py

@@ -1,6 +1,6 @@
 """IBM WatsonX embedding function implementation."""
 
-from typing import cast
+from typing import Any, cast
 
 from chromadb.api.types import Documents, EmbeddingFunction, Embeddings
 from typing_extensions import Unpack
@@ -15,14 +15,18 @@ _printer = Printer()
 class WatsonXEmbeddingFunction(EmbeddingFunction[Documents]):
     """Embedding function for IBM WatsonX models."""
 
-    def __init__(self, **kwargs: Unpack[WatsonXProviderConfig]) -> None:
+    def __init__(
+        self, *, verbose: bool = True, **kwargs: Unpack[WatsonXProviderConfig]
+    ) -> None:
         """Initialize WatsonX embedding function.
 
         Args:
+            verbose: Whether to print error messages.
             **kwargs: Configuration parameters for WatsonX Embeddings and Credentials.
         """
         super().__init__(**kwargs)
         self._config = kwargs
+        self._verbose = verbose
 
     @staticmethod
     def name() -> str:
@@ -56,7 +60,7 @@ class WatsonXEmbeddingFunction(EmbeddingFunction[Documents]):
         if isinstance(input, str):
             input = [input]
 
-        embeddings_config: dict = {
+        embeddings_config: dict[str, Any] = {
             "model_id": self._config["model_id"],
         }
         if "params" in self._config and self._config["params"] is not None:
@@ -90,7 +94,7 @@ class WatsonXEmbeddingFunction(EmbeddingFunction[Documents]):
         if "credentials" in self._config and self._config["credentials"] is not None:
             embeddings_config["credentials"] = self._config["credentials"]
         else:
-            cred_config: dict = {}
+            cred_config: dict[str, Any] = {}
             if "url" in self._config and self._config["url"] is not None:
                 cred_config["url"] = self._config["url"]
             if "api_key" in self._config and self._config["api_key"] is not None:
@@ -159,5 +163,6 @@ class WatsonXEmbeddingFunction(EmbeddingFunction[Documents]):
             embeddings = embedding.embed_documents(input)
             return cast(Embeddings, embeddings)
         except Exception as e:
-            _printer.print(f"Error during WatsonX embedding: {e}", color="red")
+            if self._verbose:
+                _printer.print(f"Error during WatsonX embedding: {e}", color="red")
             raise

lib/crewai/src/crewai/task.py

@@ -767,10 +767,11 @@ class Task(BaseModel):
             if files:
                 supported_types: list[str] = []
                 if self.agent.llm and self.agent.llm.supports_multimodal():
-                    provider = getattr(self.agent.llm, "provider", None) or getattr(
-                        self.agent.llm, "model", "openai"
+                    provider: str = str(
+                        getattr(self.agent.llm, "provider", None)
+                        or getattr(self.agent.llm, "model", "openai")
                     )
-                    api = getattr(self.agent.llm, "api", None)
+                    api: str | None = getattr(self.agent.llm, "api", None)
                     supported_types = get_supported_content_types(provider, api)
 
                 def is_auto_injected(content_type: str) -> bool:
@@ -887,10 +888,11 @@ Follow these guidelines:
             try:
                 crew_chat_messages = json.loads(crew_chat_messages_json)
             except json.JSONDecodeError as e:
-                _printer.print(
-                    f"An error occurred while parsing crew chat messages: {e}",
-                    color="red",
-                )
+                if self.agent and self.agent.verbose:
+                    _printer.print(
+                        f"An error occurred while parsing crew chat messages: {e}",
+                        color="red",
+                    )
                 raise
 
             conversation_history = "\n".join(
@@ -1132,11 +1134,12 @@ Follow these guidelines:
                 guardrail_result_error=guardrail_result.error,
                 task_output=task_output.raw,
             )
-            printer = Printer()
-            printer.print(
-                content=f"Guardrail {guardrail_index if guardrail_index is not None else ''} blocked (attempt {attempt + 1}/{max_attempts}), retrying due to: {guardrail_result.error}\n",
-                color="yellow",
-            )
+            if agent and agent.verbose:
+                printer = Printer()
+                printer.print(
+                    content=f"Guardrail {guardrail_index if guardrail_index is not None else ''} blocked (attempt {attempt + 1}/{max_attempts}), retrying due to: {guardrail_result.error}\n",
+                    color="yellow",
+                )
 
             # Regenerate output from agent
             result = agent.execute_task(
@@ -1229,11 +1232,12 @@ Follow these guidelines:
                 guardrail_result_error=guardrail_result.error,
                 task_output=task_output.raw,
             )
-            printer = Printer()
-            printer.print(
-                content=f"Guardrail {guardrail_index if guardrail_index is not None else ''} blocked (attempt {attempt + 1}/{max_attempts}), retrying due to: {guardrail_result.error}\n",
-                color="yellow",
-            )
+            if agent and agent.verbose:
+                printer = Printer()
+                printer.print(
+                    content=f"Guardrail {guardrail_index if guardrail_index is not None else ''} blocked (attempt {attempt + 1}/{max_attempts}), retrying due to: {guardrail_result.error}\n",
+                    color="yellow",
+                )
 
             result = await agent.aexecute_task(
                 task=self,

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

@@ -384,6 +384,8 @@ class ToolUsage:
                         if (
                             hasattr(available_tool, "max_usage_count")
                             and available_tool.max_usage_count is not None
+                            and self.agent
+                            and self.agent.verbose
                         ):
                             self._printer.print(
                                 content=f"Tool '{sanitize_tool_name(available_tool.name)}' usage: {available_tool.current_usage_count}/{available_tool.max_usage_count}",
@@ -396,6 +398,8 @@ class ToolUsage:
                         if (
                             hasattr(available_tool, "max_usage_count")
                             and available_tool.max_usage_count is not None
+                            and self.agent
+                            and self.agent.verbose
                         ):
                             self._printer.print(
                                 content=f"Tool '{sanitize_tool_name(available_tool.name)}' usage: {available_tool.current_usage_count}/{available_tool.max_usage_count}",
@@ -610,6 +614,8 @@ class ToolUsage:
                         if (
                             hasattr(available_tool, "max_usage_count")
                             and available_tool.max_usage_count is not None
+                            and self.agent
+                            and self.agent.verbose
                         ):
                             self._printer.print(
                                 content=f"Tool '{sanitize_tool_name(available_tool.name)}' usage: {available_tool.current_usage_count}/{available_tool.max_usage_count}",
@@ -622,6 +628,8 @@ class ToolUsage:
                         if (
                             hasattr(available_tool, "max_usage_count")
                             and available_tool.max_usage_count is not None
+                            and self.agent
+                            and self.agent.verbose
                         ):
                             self._printer.print(
                                 content=f"Tool '{sanitize_tool_name(available_tool.name)}' usage: {available_tool.current_usage_count}/{available_tool.max_usage_count}",
@@ -884,15 +892,17 @@ class ToolUsage:
         # Attempt 4: Repair JSON
         try:
             repaired_input = str(repair_json(tool_input, skip_json_loads=True))
-            self._printer.print(
-                content=f"Repaired JSON: {repaired_input}", color="blue"
-            )
+            if self.agent and self.agent.verbose:
+                self._printer.print(
+                    content=f"Repaired JSON: {repaired_input}", color="blue"
+                )
             arguments = json.loads(repaired_input)
             if isinstance(arguments, dict):
                 return arguments
         except Exception as e:
             error = f"Failed to repair JSON: {e}"
-            self._printer.print(content=error, color="red")
+            if self.agent and self.agent.verbose:
+                self._printer.print(content=error, color="red")
 
         error_message = (
             "Tool input must be a valid dictionary in JSON or Python literal format"

lib/crewai/src/crewai/translations/en.json

@@ -10,9 +10,10 @@
     "memory": "\n\n# Useful context: \n{memory}",
     "role_playing": "You are {role}. {backstory}\nYour personal goal is: {goal}",
     "tools": "\nYou ONLY have access to the following tools, and should NEVER make up tools that are not listed here:\n\n{tools}\n\nIMPORTANT: Use the following format in your response:\n\n```\nThought: you should always think about what to do\nAction: the action to take, only one name of [{tool_names}], just the name, exactly as it's written.\nAction Input: the input to the action, just a simple JSON object, enclosed in curly braces, using \" to wrap keys and values.\nObservation: the result of the action\n```\n\nOnce all necessary information is gathered, return the following format:\n\n```\nThought: I now know the final answer\nFinal Answer: the final answer to the original input question\n```",
-    "no_tools": "\nTo give my best complete final answer to the task respond using the exact following format:\n\nThought: I now can give a great answer\nFinal Answer: Your final answer must be the great and the most complete as possible, it must be outcome described.\n\nI MUST use these formats, my job depends on it!",
-    "native_tools": "\nUse available tools to gather information and complete your task.",
-    "native_task": "\nCurrent Task: {input}\n\nThis is VERY important to you, your job depends on it!",
+    "no_tools": "",
+    "task_no_tools": "\nCurrent Task: {input}\n\nProvide your complete response:",
+    "native_tools": "",
+    "native_task": "\nCurrent Task: {input}",
     "post_tool_reasoning": "Analyze the tool result. If requirements are met, provide the Final Answer. Otherwise, call the next tool. Deliver only the answer without meta-commentary.",
     "format": "Decide if you need a tool or can provide the final answer. Use one at a time.\nTo use a tool, use:\nThought: [reasoning]\nAction: [name from {tool_names}]\nAction Input: [JSON object]\n\nTo provide the final answer, use:\nThought: [reasoning]\nFinal Answer: [complete response]",
     "final_answer_format": "If you don't need to use any more tools, you must give your best complete final answer, make sure it satisfies the expected criteria, use the EXACT format below:\n\n```\nThought: I now can give a great answer\nFinal Answer: my best complete final answer to the task.\n\n```",

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

@@ -210,6 +210,7 @@ def handle_max_iterations_exceeded(
     messages: list[LLMMessage],
     llm: LLM | BaseLLM,
     callbacks: list[TokenCalcHandler],
+    verbose: bool = True,
 ) -> AgentFinish:
     """Handles the case when the maximum number of iterations is exceeded. Performs one more LLM call to get the final answer.
 
@@ -220,14 +221,16 @@ def handle_max_iterations_exceeded(
         messages: List of messages to send to the LLM.
         llm: The LLM instance to call.
         callbacks: List of callbacks for the LLM call.
+        verbose: Whether to print output.
 
     Returns:
         AgentFinish with the final answer after exceeding max iterations.
     """
-    printer.print(
-        content="Maximum iterations reached. Requesting final answer.",
-        color="yellow",
-    )
+    if verbose:
+        printer.print(
+            content="Maximum iterations reached. Requesting final answer.",
+            color="yellow",
+        )
 
     if formatted_answer and hasattr(formatted_answer, "text"):
         assistant_message = (
@@ -245,10 +248,11 @@ def handle_max_iterations_exceeded(
     )
 
     if answer is None or answer == "":
-        printer.print(
-            content="Received None or empty response from LLM call.",
-            color="red",
-        )
+        if verbose:
+            printer.print(
+                content="Received None or empty response from LLM call.",
+                color="red",
+            )
         raise ValueError("Invalid response from LLM call - None or empty.")
 
     formatted = format_answer(answer=answer)
@@ -322,7 +326,8 @@ def get_llm_response(
     from_agent: Agent | LiteAgent | None = None,
     response_model: type[BaseModel] | None = None,
     executor_context: CrewAgentExecutor | AgentExecutor | LiteAgent | None = None,
-) -> str | Any:
+    verbose: bool = True,
+) -> str | BaseModel | Any:
     """Call the LLM and return the response, handling any invalid responses.
 
     Args:
@@ -336,10 +341,11 @@ def get_llm_response(
         from_agent: Optional agent context for the LLM call.
         response_model: Optional Pydantic model for structured outputs.
         executor_context: Optional executor context for hook invocation.
+        verbose: Whether to print output.
 
     Returns:
-        The response from the LLM as a string, or tool call results if
-        native function calling is used.
+        The response from the LLM as a string, Pydantic model (when response_model is provided),
+        or tool call results if native function calling is used.
 
     Raises:
         Exception: If an error occurs.
@@ -347,7 +353,7 @@ def get_llm_response(
     """
 
     if executor_context is not None:
-        if not _setup_before_llm_call_hooks(executor_context, printer):
+        if not _setup_before_llm_call_hooks(executor_context, printer, verbose=verbose):
             raise ValueError("LLM call blocked by before_llm_call hook")
         messages = executor_context.messages
 
@@ -364,13 +370,16 @@ def get_llm_response(
     except Exception as e:
         raise e
     if not answer:
-        printer.print(
-            content="Received None or empty response from LLM call.",
-            color="red",
-        )
+        if verbose:
+            printer.print(
+                content="Received None or empty response from LLM call.",
+                color="red",
+            )
         raise ValueError("Invalid response from LLM call - None or empty.")
 
-    return _setup_after_llm_call_hooks(executor_context, answer, printer)
+    return _setup_after_llm_call_hooks(
+        executor_context, answer, printer, verbose=verbose
+    )
 
 
 async def aget_llm_response(
@@ -384,7 +393,8 @@ async def aget_llm_response(
     from_agent: Agent | LiteAgent | None = None,
     response_model: type[BaseModel] | None = None,
     executor_context: CrewAgentExecutor | AgentExecutor | None = None,
-) -> str | Any:
+    verbose: bool = True,
+) -> str | BaseModel | Any:
     """Call the LLM asynchronously and return the response.
 
     Args:
@@ -400,15 +410,15 @@ async def aget_llm_response(
         executor_context: Optional executor context for hook invocation.
 
     Returns:
-        The response from the LLM as a string, or tool call results if
-        native function calling is used.
+        The response from the LLM as a string, Pydantic model (when response_model is provided),
+        or tool call results if native function calling is used.
 
     Raises:
         Exception: If an error occurs.
         ValueError: If the response is None or empty.
     """
     if executor_context is not None:
-        if not _setup_before_llm_call_hooks(executor_context, printer):
+        if not _setup_before_llm_call_hooks(executor_context, printer, verbose=verbose):
             raise ValueError("LLM call blocked by before_llm_call hook")
         messages = executor_context.messages
 
@@ -425,13 +435,16 @@ async def aget_llm_response(
     except Exception as e:
         raise e
     if not answer:
-        printer.print(
-            content="Received None or empty response from LLM call.",
-            color="red",
-        )
+        if verbose:
+            printer.print(
+                content="Received None or empty response from LLM call.",
+                color="red",
+            )
         raise ValueError("Invalid response from LLM call - None or empty.")
 
-    return _setup_after_llm_call_hooks(executor_context, answer, printer)
+    return _setup_after_llm_call_hooks(
+        executor_context, answer, printer, verbose=verbose
+    )
 
 
 def process_llm_response(
@@ -498,13 +511,19 @@ def handle_agent_action_core(
     return formatted_answer
 
 
-def handle_unknown_error(printer: Printer, exception: Exception) -> None:
+def handle_unknown_error(
+    printer: Printer, exception: Exception, verbose: bool = True
+) -> None:
     """Handle unknown errors by informing the user.
 
     Args:
         printer: Printer instance for output
         exception: The exception that occurred
+        verbose: Whether to print output.
     """
+    if not verbose:
+        return
+
     error_message = str(exception)
 
     if "litellm" in error_message:
@@ -526,6 +545,7 @@ def handle_output_parser_exception(
     iterations: int,
     log_error_after: int = 3,
     printer: Printer | None = None,
+    verbose: bool = True,
 ) -> AgentAction:
     """Handle OutputParserError by updating messages and formatted_answer.
 
@@ -548,7 +568,7 @@ def handle_output_parser_exception(
         thought="",
     )
 
-    if iterations > log_error_after and printer:
+    if verbose and iterations > log_error_after and printer:
         printer.print(
             content=f"Error parsing LLM output, agent will retry: {e.error}",
             color="red",
@@ -578,6 +598,7 @@ def handle_context_length(
     llm: LLM | BaseLLM,
     callbacks: list[TokenCalcHandler],
     i18n: I18N,
+    verbose: bool = True,
 ) -> None:
     """Handle context length exceeded by either summarizing or raising an error.
 
@@ -593,16 +614,20 @@ def handle_context_length(
         SystemExit: If context length is exceeded and user opts not to summarize
     """
     if respect_context_window:
-        printer.print(
-            content="Context length exceeded. Summarizing content to fit the model context window. Might take a while...",
-            color="yellow",
+        if verbose:
+            printer.print(
+                content="Context length exceeded. Summarizing content to fit the model context window. Might take a while...",
+                color="yellow",
+            )
+        summarize_messages(
+            messages=messages, llm=llm, callbacks=callbacks, i18n=i18n, verbose=verbose
         )
-        summarize_messages(messages=messages, llm=llm, callbacks=callbacks, i18n=i18n)
     else:
-        printer.print(
-            content="Context length exceeded. Consider using smaller text or RAG tools from crewai_tools.",
-            color="red",
-        )
+        if verbose:
+            printer.print(
+                content="Context length exceeded. Consider using smaller text or RAG tools from crewai_tools.",
+                color="red",
+            )
         raise SystemExit(
             "Context length exceeded and user opted not to summarize. Consider using smaller text or RAG tools from crewai_tools."
         )
@@ -613,6 +638,7 @@ def summarize_messages(
     llm: LLM | BaseLLM,
     callbacks: list[TokenCalcHandler],
     i18n: I18N,
+    verbose: bool = True,
 ) -> None:
     """Summarize messages to fit within context window.
 
@@ -644,10 +670,11 @@ def summarize_messages(
 
     total_groups = len(messages_groups)
     for idx, group in enumerate(messages_groups, 1):
-        Printer().print(
-            content=f"Summarizing {idx}/{total_groups}...",
-            color="yellow",
-        )
+        if verbose:
+            Printer().print(
+                content=f"Summarizing {idx}/{total_groups}...",
+                color="yellow",
+            )
 
         summarization_messages = [
             format_message_for_llm(
@@ -905,12 +932,14 @@ def extract_tool_call_info(
 def _setup_before_llm_call_hooks(
     executor_context: CrewAgentExecutor | AgentExecutor | LiteAgent | None,
     printer: Printer,
+    verbose: bool = True,
 ) -> bool:
     """Setup and invoke before_llm_call hooks for the executor context.
 
     Args:
         executor_context: The executor context to setup the hooks for.
         printer: Printer instance for error logging.
+        verbose: Whether to print output.
 
     Returns:
         True if LLM execution should proceed, False if blocked by a hook.
@@ -925,26 +954,29 @@ def _setup_before_llm_call_hooks(
             for hook in executor_context.before_llm_call_hooks:
                 result = hook(hook_context)
                 if result is False:
-                    printer.print(
-                        content="LLM call blocked by before_llm_call hook",
-                        color="yellow",
-                    )
+                    if verbose:
+                        printer.print(
+                            content="LLM call blocked by before_llm_call hook",
+                            color="yellow",
+                        )
                     return False
         except Exception as e:
-            printer.print(
-                content=f"Error in before_llm_call hook: {e}",
-                color="yellow",
-            )
+            if verbose:
+                printer.print(
+                    content=f"Error in before_llm_call hook: {e}",
+                    color="yellow",
+                )
 
         if not isinstance(executor_context.messages, list):
-            printer.print(
-                content=(
-                    "Warning: before_llm_call hook replaced messages with non-list. "
-                    "Restoring original messages list. Hooks should modify messages in-place, "
-                    "not replace the list (e.g., use context.messages.append() not context.messages = [])."
-                ),
-                color="yellow",
-            )
+            if verbose:
+                printer.print(
+                    content=(
+                        "Warning: before_llm_call hook replaced messages with non-list. "
+                        "Restoring original messages list. Hooks should modify messages in-place, "
+                        "not replace the list (e.g., use context.messages.append() not context.messages = [])."
+                    ),
+                    color="yellow",
+                )
             if isinstance(original_messages, list):
                 executor_context.messages = original_messages
             else:
@@ -955,49 +987,79 @@ def _setup_before_llm_call_hooks(
 
 def _setup_after_llm_call_hooks(
     executor_context: CrewAgentExecutor | AgentExecutor | LiteAgent | None,
-    answer: str,
+    answer: str | BaseModel,
     printer: Printer,
-) -> str:
+    verbose: bool = True,
+) -> str | BaseModel:
     """Setup and invoke after_llm_call hooks for the executor context.
 
     Args:
         executor_context: The executor context to setup the hooks for.
-        answer: The LLM response string.
+        answer: The LLM response (string or Pydantic model).
         printer: Printer instance for error logging.
+        verbose: Whether to print output.
 
     Returns:
-        The potentially modified response string.
+        The potentially modified response (string or Pydantic model).
     """
     if executor_context and executor_context.after_llm_call_hooks:
         from crewai.hooks.llm_hooks import LLMCallHookContext
 
         original_messages = executor_context.messages
 
-        hook_context = LLMCallHookContext(executor_context, response=answer)
+        # For Pydantic models, serialize to JSON for hooks
+        if isinstance(answer, BaseModel):
+            pydantic_answer = answer
+            hook_response: str = pydantic_answer.model_dump_json()
+            original_json: str = hook_response
+        else:
+            pydantic_answer = None
+            hook_response = str(answer)
+
+        hook_context = LLMCallHookContext(executor_context, response=hook_response)
         try:
             for hook in executor_context.after_llm_call_hooks:
                 modified_response = hook(hook_context)
                 if modified_response is not None and isinstance(modified_response, str):
-                    answer = modified_response
+                    hook_response = modified_response
 
         except Exception as e:
-            printer.print(
-                content=f"Error in after_llm_call hook: {e}",
-                color="yellow",
-            )
+            if verbose:
+                printer.print(
+                    content=f"Error in after_llm_call hook: {e}",
+                    color="yellow",
+                )
 
         if not isinstance(executor_context.messages, list):
-            printer.print(
-                content=(
-                    "Warning: after_llm_call hook replaced messages with non-list. "
-                    "Restoring original messages list. Hooks should modify messages in-place, "
-                    "not replace the list (e.g., use context.messages.append() not context.messages = [])."
-                ),
-                color="yellow",
-            )
+            if verbose:
+                printer.print(
+                    content=(
+                        "Warning: after_llm_call hook replaced messages with non-list. "
+                        "Restoring original messages list. Hooks should modify messages in-place, "
+                        "not replace the list (e.g., use context.messages.append() not context.messages = [])."
+                    ),
+                    color="yellow",
+                )
             if isinstance(original_messages, list):
                 executor_context.messages = original_messages
             else:
                 executor_context.messages = []
 
+        # If hooks modified the response, update answer accordingly
+        if pydantic_answer is not None:
+            # For Pydantic models, reparse the JSON if it was modified
+            if hook_response != original_json:
+                try:
+                    model_class: type[BaseModel] = type(pydantic_answer)
+                    answer = model_class.model_validate_json(hook_response)
+                except Exception as e:
+                    if verbose:
+                        printer.print(
+                            content=f"Warning: Hook modified response but failed to reparse as {type(pydantic_answer).__name__}: {e}. Using original model.",
+                            color="yellow",
+                        )
+        else:
+            # For string responses, use the hook-modified response
+            answer = hook_response
+
     return answer

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

@@ -62,7 +62,10 @@ class Converter(OutputConverter):
                     ],
                     response_model=self.model,
                 )
-                result = self.model.model_validate_json(response)
+                if isinstance(response, BaseModel):
+                    result = response
+                else:
+                    result = self.model.model_validate_json(response)
             else:
                 response = self.llm.call(
                     [
@@ -205,10 +208,11 @@ def convert_to_model(
         )
 
     except Exception as e:
-        Printer().print(
-            content=f"Unexpected error during model conversion: {type(e).__name__}: {e}. Returning original result.",
-            color="red",
-        )
+        if agent and getattr(agent, "verbose", True):
+            Printer().print(
+                content=f"Unexpected error during model conversion: {type(e).__name__}: {e}. Returning original result.",
+                color="red",
+            )
         return result
 
 
@@ -262,10 +266,11 @@ def handle_partial_json(
         except ValidationError:
             raise
         except Exception as e:
-            Printer().print(
-                content=f"Unexpected error during partial JSON handling: {type(e).__name__}: {e}. Attempting alternative conversion method.",
-                color="red",
-            )
+            if agent and getattr(agent, "verbose", True):
+                Printer().print(
+                    content=f"Unexpected error during partial JSON handling: {type(e).__name__}: {e}. Attempting alternative conversion method.",
+                    color="red",
+                )
 
     return convert_with_instructions(
         result=result,
@@ -323,10 +328,11 @@ def convert_with_instructions(
     )
 
     if isinstance(exported_result, ConverterError):
-        Printer().print(
-            content=f"Failed to convert result to model: {exported_result}",
-            color="red",
-        )
+        if agent and getattr(agent, "verbose", True):
+            Printer().print(
+                content=f"Failed to convert result to model: {exported_result}",
+                color="red",
+            )
         return result
 
     return exported_result

lib/crewai/src/crewai/utilities/prompts.py

@@ -23,7 +23,13 @@ class SystemPromptResult(StandardPromptResult):
 
 
 COMPONENTS = Literal[
-    "role_playing", "tools", "no_tools", "native_tools", "task", "native_task"
+    "role_playing",
+    "tools",
+    "no_tools",
+    "native_tools",
+    "task",
+    "native_task",
+    "task_no_tools",
 ]
 
 
@@ -74,11 +80,14 @@ class Prompts(BaseModel):
             slices.append("no_tools")
         system: str = self._build_prompt(slices)
 
-        # Use native_task for native tool calling (no "Thought:" prompt)
-        # Use task for ReAct pattern (includes "Thought:" prompt)
-        task_slice: COMPONENTS = (
-            "native_task" if self.use_native_tool_calling else "task"
-        )
+        # Determine which task slice to use:
+        task_slice: COMPONENTS
+        if self.use_native_tool_calling:
+            task_slice = "native_task"
+        elif self.has_tools:
+            task_slice = "task"
+        else:
+            task_slice = "task_no_tools"
         slices.append(task_slice)
 
         if (

lib/crewai/tests/agents/test_lite_agent.py

@@ -1004,3 +1004,53 @@ def test_prepare_kickoff_param_files_override_message_files():
 
     assert "files" in inputs
     assert inputs["files"]["same.png"] is param_file  # param takes precedence
+
+
+def test_lite_agent_verbose_false_suppresses_printer_output():
+    """Test that setting verbose=False suppresses all printer output."""
+    from crewai.agents.parser import AgentFinish
+    from crewai.types.usage_metrics import UsageMetrics
+
+    mock_llm = Mock(spec=LLM)
+    mock_llm.call.return_value = "Final Answer: Hello!"
+    mock_llm.stop = []
+    mock_llm.supports_stop_words.return_value = False
+    mock_llm.get_token_usage_summary.return_value = UsageMetrics(
+        total_tokens=100,
+        prompt_tokens=50,
+        completion_tokens=50,
+        cached_prompt_tokens=0,
+        successful_requests=1,
+    )
+
+    with pytest.warns(DeprecationWarning):
+        agent = LiteAgent(
+            role="Test Agent",
+            goal="Test goal",
+            backstory="Test backstory",
+            llm=mock_llm,
+            verbose=False,
+        )
+
+    result = agent.kickoff("Say hello")
+
+    assert result is not None
+    assert isinstance(result, LiteAgentOutput)
+    # Verify the printer was never called
+    agent._printer.print = Mock()
+    # For a clean verification, patch printer before execution
+    with pytest.warns(DeprecationWarning):
+        agent2 = LiteAgent(
+            role="Test Agent",
+            goal="Test goal",
+            backstory="Test backstory",
+            llm=mock_llm,
+            verbose=False,
+        )
+
+    mock_printer = Mock()
+    agent2._printer = mock_printer
+
+    agent2.kickoff("Say hello")
+
+    mock_printer.print.assert_not_called()

lib/crewai/tests/cassettes/utilities/TestRealLLMNoThoughtLeakage.test_agent_without_tools_no_thought_in_output.yaml

@@ -0,0 +1,112 @@
+interactions:
+- request:
+    body: '{"messages":[{"role":"system","content":"You are Language Detector. You
+      are an expert linguist who can identify languages.\nYour personal goal is: Detect
+      the language of text"},{"role":"user","content":"\nCurrent Task: What language
+      is this text written in: ''Hello, how are you?''\n\nThis is the expected criteria
+      for your final answer: The detected language (e.g., English, Spanish, etc.)\nyou
+      MUST return the actual complete content as the final answer, not a summary.\n\nProvide
+      your complete response:"}],"model":"gpt-4o-mini"}'
+    headers:
+      User-Agent:
+      - X-USER-AGENT-XXX
+      accept:
+      - application/json
+      accept-encoding:
+      - ACCEPT-ENCODING-XXX
+      authorization:
+      - AUTHORIZATION-XXX
+      connection:
+      - keep-alive
+      content-length:
+      - '530'
+      content-type:
+      - application/json
+      host:
+      - api.openai.com
+      x-stainless-arch:
+      - X-STAINLESS-ARCH-XXX
+      x-stainless-async:
+      - 'false'
+      x-stainless-lang:
+      - python
+      x-stainless-os:
+      - X-STAINLESS-OS-XXX
+      x-stainless-package-version:
+      - 1.83.0
+      x-stainless-read-timeout:
+      - X-STAINLESS-READ-TIMEOUT-XXX
+      x-stainless-retry-count:
+      - '0'
+      x-stainless-runtime:
+      - CPython
+      x-stainless-runtime-version:
+      - 3.13.3
+    method: POST
+    uri: https://api.openai.com/v1/chat/completions
+  response:
+    body:
+      string: "{\n  \"id\": \"chatcmpl-D39bkotgEapBcz1sSIXvhPhK9G7FD\",\n  \"object\":
+        \"chat.completion\",\n  \"created\": 1769644288,\n  \"model\": \"gpt-4o-mini-2024-07-18\",\n
+        \ \"choices\": [\n    {\n      \"index\": 0,\n      \"message\": {\n        \"role\":
+        \"assistant\",\n        \"content\": \"English\",\n        \"refusal\": null,\n
+        \       \"annotations\": []\n      },\n      \"logprobs\": null,\n      \"finish_reason\":
+        \"stop\"\n    }\n  ],\n  \"usage\": {\n    \"prompt_tokens\": 101,\n    \"completion_tokens\":
+        1,\n    \"total_tokens\": 102,\n    \"prompt_tokens_details\": {\n      \"cached_tokens\":
+        0,\n      \"audio_tokens\": 0\n    },\n    \"completion_tokens_details\":
+        {\n      \"reasoning_tokens\": 0,\n      \"audio_tokens\": 0,\n      \"accepted_prediction_tokens\":
+        0,\n      \"rejected_prediction_tokens\": 0\n    }\n  },\n  \"service_tier\":
+        \"default\",\n  \"system_fingerprint\": \"fp_3683ee3deb\"\n}\n"
+    headers:
+      CF-RAY:
+      - CF-RAY-XXX
+      Connection:
+      - keep-alive
+      Content-Type:
+      - application/json
+      Date:
+      - Wed, 28 Jan 2026 23:51:28 GMT
+      Server:
+      - cloudflare
+      Set-Cookie:
+      - SET-COOKIE-XXX
+      Strict-Transport-Security:
+      - STS-XXX
+      Transfer-Encoding:
+      - chunked
+      X-Content-Type-Options:
+      - X-CONTENT-TYPE-XXX
+      access-control-expose-headers:
+      - ACCESS-CONTROL-XXX
+      alt-svc:
+      - h3=":443"; ma=86400
+      cf-cache-status:
+      - DYNAMIC
+      openai-organization:
+      - OPENAI-ORG-XXX
+      openai-processing-ms:
+      - '279'
+      openai-project:
+      - OPENAI-PROJECT-XXX
+      openai-version:
+      - '2020-10-01'
+      x-openai-proxy-wasm:
+      - v0.1
+      x-ratelimit-limit-requests:
+      - X-RATELIMIT-LIMIT-REQUESTS-XXX
+      x-ratelimit-limit-tokens:
+      - X-RATELIMIT-LIMIT-TOKENS-XXX
+      x-ratelimit-remaining-requests:
+      - X-RATELIMIT-REMAINING-REQUESTS-XXX
+      x-ratelimit-remaining-tokens:
+      - X-RATELIMIT-REMAINING-TOKENS-XXX
+      x-ratelimit-reset-requests:
+      - X-RATELIMIT-RESET-REQUESTS-XXX
+      x-ratelimit-reset-tokens:
+      - X-RATELIMIT-RESET-TOKENS-XXX
+      x-request-id:
+      - X-REQUEST-ID-XXX
+    status:
+      code: 200
+      message: OK
+version: 1

lib/crewai/tests/cassettes/utilities/TestRealLLMNoThoughtLeakage.test_simple_task_clean_output.yaml

@@ -0,0 +1,111 @@
+interactions:
+- request:
+    body: '{"messages":[{"role":"system","content":"You are Classifier. You classify
+      text sentiment accurately.\nYour personal goal is: Classify text sentiment"},{"role":"user","content":"\nCurrent
+      Task: Classify the sentiment of: ''I love this product!''\n\nThis is the expected
+      criteria for your final answer: One word: positive, negative, or neutral\nyou
+      MUST return the actual complete content as the final answer, not a summary.\n\nProvide
+      your complete response:"}],"model":"gpt-4o-mini"}'
+    headers:
+      User-Agent:
+      - X-USER-AGENT-XXX
+      accept:
+      - application/json
+      accept-encoding:
+      - ACCEPT-ENCODING-XXX
+      authorization:
+      - AUTHORIZATION-XXX
+      connection:
+      - keep-alive
+      content-length:
+      - '481'
+      content-type:
+      - application/json
+      host:
+      - api.openai.com
+      x-stainless-arch:
+      - X-STAINLESS-ARCH-XXX
+      x-stainless-async:
+      - 'false'
+      x-stainless-lang:
+      - python
+      x-stainless-os:
+      - X-STAINLESS-OS-XXX
+      x-stainless-package-version:
+      - 1.83.0
+      x-stainless-read-timeout:
+      - X-STAINLESS-READ-TIMEOUT-XXX
+      x-stainless-retry-count:
+      - '0'
+      x-stainless-runtime:
+      - CPython
+      x-stainless-runtime-version:
+      - 3.13.3
+    method: POST
+    uri: https://api.openai.com/v1/chat/completions
+  response:
+    body:
+      string: "{\n  \"id\": \"chatcmpl-D39bkVPelOZanWIMBoIyzsuj072sM\",\n  \"object\":
+        \"chat.completion\",\n  \"created\": 1769644288,\n  \"model\": \"gpt-4o-mini-2024-07-18\",\n
+        \ \"choices\": [\n    {\n      \"index\": 0,\n      \"message\": {\n        \"role\":
+        \"assistant\",\n        \"content\": \"positive\",\n        \"refusal\": null,\n
+        \       \"annotations\": []\n      },\n      \"logprobs\": null,\n      \"finish_reason\":
+        \"stop\"\n    }\n  ],\n  \"usage\": {\n    \"prompt_tokens\": 89,\n    \"completion_tokens\":
+        1,\n    \"total_tokens\": 90,\n    \"prompt_tokens_details\": {\n      \"cached_tokens\":
+        0,\n      \"audio_tokens\": 0\n    },\n    \"completion_tokens_details\":
+        {\n      \"reasoning_tokens\": 0,\n      \"audio_tokens\": 0,\n      \"accepted_prediction_tokens\":
+        0,\n      \"rejected_prediction_tokens\": 0\n    }\n  },\n  \"service_tier\":
+        \"default\",\n  \"system_fingerprint\": \"fp_3683ee3deb\"\n}\n"
+    headers:
+      CF-RAY:
+      - CF-RAY-XXX
+      Connection:
+      - keep-alive
+      Content-Type:
+      - application/json
+      Date:
+      - Wed, 28 Jan 2026 23:51:29 GMT
+      Server:
+      - cloudflare
+      Set-Cookie:
+      - SET-COOKIE-XXX
+      Strict-Transport-Security:
+      - STS-XXX
+      Transfer-Encoding:
+      - chunked
+      X-Content-Type-Options:
+      - X-CONTENT-TYPE-XXX
+      access-control-expose-headers:
+      - ACCESS-CONTROL-XXX
+      alt-svc:
+      - h3=":443"; ma=86400
+      cf-cache-status:
+      - DYNAMIC
+      openai-organization:
+      - OPENAI-ORG-XXX
+      openai-processing-ms:
+      - '323'
+      openai-project:
+      - OPENAI-PROJECT-XXX
+      openai-version:
+      - '2020-10-01'
+      x-openai-proxy-wasm:
+      - v0.1
+      x-ratelimit-limit-requests:
+      - X-RATELIMIT-LIMIT-REQUESTS-XXX
+      x-ratelimit-limit-tokens:
+      - X-RATELIMIT-LIMIT-TOKENS-XXX
+      x-ratelimit-remaining-requests:
+      - X-RATELIMIT-REMAINING-REQUESTS-XXX
+      x-ratelimit-remaining-tokens:
+      - X-RATELIMIT-REMAINING-TOKENS-XXX
+      x-ratelimit-reset-requests:
+      - X-RATELIMIT-RESET-REQUESTS-XXX
+      x-ratelimit-reset-tokens:
+      - X-RATELIMIT-RESET-TOKENS-XXX
+      x-request-id:
+      - X-REQUEST-ID-XXX
+    status:
+      code: 200
+      message: OK
+version: 1

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

@@ -157,10 +157,10 @@ async def test_anthropic_async_with_response_model():
         "Say hello in French",
         response_model=GreetingResponse
     )
-    model = GreetingResponse.model_validate_json(result)
-    assert isinstance(model, GreetingResponse)
-    assert isinstance(model.greeting, str)
-    assert isinstance(model.language, str)
+    # When response_model is provided, the result is already a parsed Pydantic model instance
+    assert isinstance(result, GreetingResponse)
+    assert isinstance(result.greeting, str)
+    assert isinstance(result.language, str)
 
 
 @pytest.mark.vcr()

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

@@ -799,3 +799,131 @@ def test_google_express_mode_works() -> None:
         assert result.token_usage.prompt_tokens > 0
         assert result.token_usage.completion_tokens > 0
         assert result.token_usage.successful_requests >= 1
+
+
+def test_gemini_2_0_model_detection():
+    """Test that Gemini 2.0 models are properly detected."""
+    # Test Gemini 2.0 models
+    llm_2_0 = LLM(model="google/gemini-2.0-flash-001")
+    from crewai.llms.providers.gemini.completion import GeminiCompletion
+    assert isinstance(llm_2_0, GeminiCompletion)
+    assert llm_2_0.is_gemini_2_0 is True
+
+    llm_2_5 = LLM(model="google/gemini-2.5-flash")
+    assert isinstance(llm_2_5, GeminiCompletion)
+    assert llm_2_5.is_gemini_2_0 is True
+
+    # Test non-2.0 models
+    llm_1_5 = LLM(model="google/gemini-1.5-pro")
+    assert isinstance(llm_1_5, GeminiCompletion)
+    assert llm_1_5.is_gemini_2_0 is False
+
+
+def test_add_property_ordering_to_schema():
+    """Test that _add_property_ordering correctly adds propertyOrdering to schemas."""
+    from crewai.llms.providers.gemini.completion import GeminiCompletion
+
+    # Test simple object schema
+    simple_schema = {
+        "type": "object",
+        "properties": {
+            "name": {"type": "string"},
+            "age": {"type": "integer"},
+            "email": {"type": "string"}
+        }
+    }
+
+    result = GeminiCompletion._add_property_ordering(simple_schema)
+
+    assert "propertyOrdering" in result
+    assert result["propertyOrdering"] == ["name", "age", "email"]
+
+    # Test nested object schema
+    nested_schema = {
+        "type": "object",
+        "properties": {
+            "user": {
+                "type": "object",
+                "properties": {
+                    "name": {"type": "string"},
+                    "contact": {
+                        "type": "object",
+                        "properties": {
+                            "email": {"type": "string"},
+                            "phone": {"type": "string"}
+                        }
+                    }
+                }
+            },
+            "id": {"type": "integer"}
+        }
+    }
+
+    result = GeminiCompletion._add_property_ordering(nested_schema)
+
+    assert "propertyOrdering" in result
+    assert result["propertyOrdering"] == ["user", "id"]
+    assert "propertyOrdering" in result["properties"]["user"]
+    assert result["properties"]["user"]["propertyOrdering"] == ["name", "contact"]
+    assert "propertyOrdering" in result["properties"]["user"]["properties"]["contact"]
+    assert result["properties"]["user"]["properties"]["contact"]["propertyOrdering"] == ["email", "phone"]
+
+
+def test_gemini_2_0_response_model_with_property_ordering():
+    """Test that Gemini 2.0 models include propertyOrdering in response schemas."""
+    from pydantic import BaseModel, Field
+
+    class TestResponse(BaseModel):
+        """Test response model."""
+        name: str = Field(..., description="The name")
+        age: int = Field(..., description="The age")
+        email: str = Field(..., description="The email")
+
+    llm = LLM(model="google/gemini-2.0-flash-001")
+
+    # Prepare generation config with response model
+    config = llm._prepare_generation_config(response_model=TestResponse)
+
+    # Verify that the config has response_json_schema
+    assert hasattr(config, 'response_json_schema') or 'response_json_schema' in config.__dict__
+
+    # Get the schema
+    if hasattr(config, 'response_json_schema'):
+        schema = config.response_json_schema
+    else:
+        schema = config.__dict__.get('response_json_schema', {})
+
+    # Verify propertyOrdering is present for Gemini 2.0
+    assert "propertyOrdering" in schema
+    assert "name" in schema["propertyOrdering"]
+    assert "age" in schema["propertyOrdering"]
+    assert "email" in schema["propertyOrdering"]
+
+
+def test_gemini_1_5_response_model_uses_response_schema():
+    """Test that Gemini 1.5 models use response_schema parameter (not response_json_schema)."""
+    from pydantic import BaseModel, Field
+
+    class TestResponse(BaseModel):
+        """Test response model."""
+        name: str = Field(..., description="The name")
+        age: int = Field(..., description="The age")
+
+    llm = LLM(model="google/gemini-1.5-pro")
+
+    # Prepare generation config with response model
+    config = llm._prepare_generation_config(response_model=TestResponse)
+
+    # Verify that the config uses response_schema (not response_json_schema)
+    assert hasattr(config, 'response_schema') or 'response_schema' in config.__dict__
+    assert not (hasattr(config, 'response_json_schema') and config.response_json_schema is not None)
+
+    # Get the schema
+    if hasattr(config, 'response_schema'):
+        schema = config.response_schema
+    else:
+        schema = config.__dict__.get('response_schema')
+
+    # For Gemini 1.5, response_schema should be the Pydantic model itself
+    # The SDK handles conversion internally
+    assert schema is TestResponse or isinstance(schema, type)

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

@@ -540,7 +540,9 @@ def test_openai_streaming_with_response_model():
         result = llm.call("Test question", response_model=TestResponse)
 
         assert result is not None
-        assert isinstance(result, str)
+        assert isinstance(result, TestResponse)
+        assert result.answer == "test"
+        assert result.confidence == 0.95
 
         assert mock_stream.called
         call_kwargs = mock_stream.call_args[1]
@@ -1395,3 +1397,56 @@ def test_openai_responses_api_both_auto_chains_work_together():
     assert params.get("previous_response_id") == "resp_123"
     assert "reasoning.encrypted_content" in params["include"]
     assert len(params["input"]) == 2  # Reasoning item + message
+
+
+def test_openai_sdk_imports_compatibility():
+    """
+    Test that all OpenAI SDK imports used by CrewAI are available.
+
+    This test verifies that the OpenAI SDK version installed provides all the
+    types and classes that CrewAI depends on. If this test fails after updating
+    the OpenAI SDK, it indicates a breaking change in the SDK that needs to be
+    addressed.
+
+    Related to issue #4300: Dependency constraints in pyproject.toml are overly strict
+    """
+    from openai import APIConnectionError, AsyncOpenAI, NotFoundError, OpenAI, Stream
+    from openai.lib.streaming.chat import ChatCompletionStream
+    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 openai.types.responses import Response
+
+    assert OpenAI is not None
+    assert AsyncOpenAI is not None
+    assert Stream is not None
+    assert APIConnectionError is not None
+    assert NotFoundError is not None
+    assert ChatCompletionStream is not None
+    assert ChatCompletion is not None
+    assert ChatCompletionChunk is not None
+    assert Choice is not None
+    assert ChoiceDelta is not None
+    assert Response is not None
+
+
+def test_openai_sdk_client_instantiation():
+    """
+    Test that OpenAI client can be instantiated with the current SDK version.
+
+    This test verifies that the OpenAI client initialization works correctly
+    with the installed SDK version, ensuring compatibility with newer versions.
+
+    Related to issue #4300: Dependency constraints in pyproject.toml are overly strict
+    """
+    from openai import AsyncOpenAI, OpenAI
+
+    client = OpenAI(api_key="test-key")
+    async_client = AsyncOpenAI(api_key="test-key")
+
+    assert client is not None
+    assert async_client is not None
+    assert hasattr(client, "chat")
+    assert hasattr(client.chat, "completions")
+    assert hasattr(async_client, "chat")
+    assert hasattr(async_client.chat, "completions")

lib/crewai/tests/test_crew.py

@@ -2585,6 +2585,7 @@ def test_warning_long_term_memory_without_entity_memory():
         goal="You research about math.",
         backstory="You're an expert in research and you love to learn new things.",
         allow_delegation=False,
+        verbose=True,
     )
 
     task1 = Task(

lib/crewai/tests/utilities/test_prompts_no_thought_leakage.py

@@ -0,0 +1,234 @@
+"""Tests for prompt generation to prevent thought leakage.
+
+These tests verify that:
+1. Agents without tools don't get ReAct format instructions
+2. The generated prompts don't encourage "Thought:" prefixes that leak into output
+3. Real LLM calls produce clean output without internal reasoning
+"""
+
+from __future__ import annotations
+
+from unittest.mock import MagicMock
+
+import pytest
+
+from crewai import Agent, Crew, Task
+from crewai.llm import LLM
+from crewai.utilities.prompts import Prompts
+
+
+class TestNoToolsPromptGeneration:
+    """Tests for prompt generation when agent has no tools."""
+
+    def test_no_tools_uses_task_no_tools_slice(self) -> None:
+        """Test that agents without tools use task_no_tools slice instead of task."""
+        mock_agent = MagicMock()
+        mock_agent.role = "Test Agent"
+        mock_agent.goal = "Test goal"
+        mock_agent.backstory = "Test backstory"
+
+        prompts = Prompts(
+            has_tools=False,
+            use_native_tool_calling=False,
+            use_system_prompt=True,
+            agent=mock_agent,
+        )
+
+        result = prompts.task_execution()
+
+        # Verify it's a SystemPromptResult with system and user keys
+        assert "system" in result
+        assert "user" in result
+        assert "prompt" in result
+
+        # The user prompt should NOT contain "Thought:" (ReAct format)
+        assert "Thought:" not in result["user"]
+
+        # The user prompt should NOT mention tools
+        assert "use the tools available" not in result["user"]
+        assert "tools available" not in result["user"].lower()
+
+        # The system prompt should NOT contain ReAct format instructions
+        assert "Thought:" not in result["system"]
+        assert "Final Answer:" not in result["system"]
+
+    def test_no_tools_prompt_is_simple(self) -> None:
+        """Test that no-tools prompt is simple and direct."""
+        mock_agent = MagicMock()
+        mock_agent.role = "Language Detector"
+        mock_agent.goal = "Detect language"
+        mock_agent.backstory = "Expert linguist"
+
+        prompts = Prompts(
+            has_tools=False,
+            use_native_tool_calling=False,
+            use_system_prompt=True,
+            agent=mock_agent,
+        )
+
+        result = prompts.task_execution()
+
+        # Should contain the role playing info
+        assert "Language Detector" in result["system"]
+
+        # User prompt should be simple with just the task
+        assert "Current Task:" in result["user"]
+        assert "Provide your complete response:" in result["user"]
+
+    def test_with_tools_uses_task_slice_with_react(self) -> None:
+        """Test that agents WITH tools use the task slice (ReAct format)."""
+        mock_agent = MagicMock()
+        mock_agent.role = "Test Agent"
+        mock_agent.goal = "Test goal"
+        mock_agent.backstory = "Test backstory"
+
+        prompts = Prompts(
+            has_tools=True,
+            use_native_tool_calling=False,
+            use_system_prompt=True,
+            agent=mock_agent,
+        )
+
+        result = prompts.task_execution()
+
+        # With tools and ReAct, the prompt SHOULD contain Thought:
+        assert "Thought:" in result["user"]
+
+    def test_native_tools_uses_native_task_slice(self) -> None:
+        """Test that native tool calling uses native_task slice."""
+        mock_agent = MagicMock()
+        mock_agent.role = "Test Agent"
+        mock_agent.goal = "Test goal"
+        mock_agent.backstory = "Test backstory"
+
+        prompts = Prompts(
+            has_tools=True,
+            use_native_tool_calling=True,
+            use_system_prompt=True,
+            agent=mock_agent,
+        )
+
+        result = prompts.task_execution()
+
+        # Native tool calling should NOT have Thought: in user prompt
+        assert "Thought:" not in result["user"]
+
+        # Should NOT have emotional manipulation
+        assert "your job depends on it" not in result["user"]
+
+
+class TestNoThoughtLeakagePatterns:
+    """Tests to verify prompts don't encourage thought leakage."""
+
+    def test_no_job_depends_on_it_in_no_tools(self) -> None:
+        """Test that 'your job depends on it' is not in no-tools prompts."""
+        mock_agent = MagicMock()
+        mock_agent.role = "Test"
+        mock_agent.goal = "Test"
+        mock_agent.backstory = "Test"
+
+        prompts = Prompts(
+            has_tools=False,
+            use_native_tool_calling=False,
+            use_system_prompt=True,
+            agent=mock_agent,
+        )
+
+        result = prompts.task_execution()
+
+        full_prompt = result["prompt"]
+        assert "your job depends on it" not in full_prompt.lower()
+        assert "i must use these formats" not in full_prompt.lower()
+
+    def test_no_job_depends_on_it_in_native_task(self) -> None:
+        """Test that 'your job depends on it' is not in native task prompts."""
+        mock_agent = MagicMock()
+        mock_agent.role = "Test"
+        mock_agent.goal = "Test"
+        mock_agent.backstory = "Test"
+
+        prompts = Prompts(
+            has_tools=True,
+            use_native_tool_calling=True,
+            use_system_prompt=True,
+            agent=mock_agent,
+        )
+
+        result = prompts.task_execution()
+
+        full_prompt = result["prompt"]
+        assert "your job depends on it" not in full_prompt.lower()
+
+
+class TestRealLLMNoThoughtLeakage:
+    """Integration tests with real LLM calls to verify no thought leakage."""
+
+    @pytest.mark.vcr()
+    def test_agent_without_tools_no_thought_in_output(self) -> None:
+        """Test that agent without tools produces clean output without 'Thought:' prefix."""
+        agent = Agent(
+            role="Language Detector",
+            goal="Detect the language of text",
+            backstory="You are an expert linguist who can identify languages.",
+            tools=[],  # No tools
+            llm=LLM(model="gpt-4o-mini"),
+            verbose=False,
+        )
+
+        task = Task(
+            description="What language is this text written in: 'Hello, how are you?'",
+            expected_output="The detected language (e.g., English, Spanish, etc.)",
+            agent=agent,
+        )
+
+        crew = Crew(agents=[agent], tasks=[task])
+        result = crew.kickoff()
+
+        assert result is not None
+        assert result.raw is not None
+
+        # The output should NOT start with "Thought:" or contain ReAct artifacts
+        output = str(result.raw)
+        assert not output.strip().startswith("Thought:")
+        assert "Final Answer:" not in output
+        assert "I now can give a great answer" not in output
+
+        # Should contain an actual answer about the language
+        assert any(
+            lang in output.lower()
+            for lang in ["english", "en", "language"]
+        )
+
+    @pytest.mark.vcr()
+    def test_simple_task_clean_output(self) -> None:
+        """Test that a simple task produces clean output without internal reasoning."""
+        agent = Agent(
+            role="Classifier",
+            goal="Classify text sentiment",
+            backstory="You classify text sentiment accurately.",
+            tools=[],
+            llm=LLM(model="gpt-4o-mini"),
+            verbose=False,
+        )
+
+        task = Task(
+            description="Classify the sentiment of: 'I love this product!'",
+            expected_output="One word: positive, negative, or neutral",
+            agent=agent,
+        )
+
+        crew = Crew(agents=[agent], tasks=[task])
+        result = crew.kickoff()
+
+        assert result is not None
+        output = str(result.raw).strip().lower()
+
+        # Output should be clean - just the classification
+        assert not output.startswith("thought:")
+        assert "final answer:" not in output
+
+        # Should contain the actual classification
+        assert any(
+            sentiment in output
+            for sentiment in ["positive", "negative", "neutral"]
+        )

lib/devtools/src/crewai_devtools/__init__.py

@@ -1,3 +1,3 @@
 """CrewAI development tools."""
 
-__version__ = "1.9.1"
+__version__ = "1.9.2"