Gemini 3 Pro support and cross-model conversation compatibility (openai#2158)

Author: ihower

pyproject.toml

@@ -37,7 +37,7 @@ Repository = "https://github.com/openai/openai-agents-python"
 [project.optional-dependencies]
 voice = ["numpy>=2.2.0, <3; python_version>='3.10'", "websockets>=15.0, <16"]
 viz = ["graphviz>=0.17"]
-litellm = ["litellm>=1.67.4.post1, <2"]
+litellm = ["litellm>=1.80.8, <2"]
 realtime = ["websockets>=15.0, <16"]
 sqlalchemy = ["SQLAlchemy>=2.0", "asyncpg>=0.29.0"]
 encrypt = ["cryptography>=45.0, <46"]

src/agents/extensions/models/litellm_model.py

@@ -62,6 +62,15 @@ class InternalChatCompletionMessage(ChatCompletionMessage):
     thinking_blocks: list[dict[str, Any]] | None = None
 
 
+class InternalToolCall(ChatCompletionMessageFunctionToolCall):
+    """
+    An internal subclass to carry provider-specific metadata (e.g., Gemini thought signatures)
+    without modifying the original model.
+    """
+
+    extra_content: dict[str, Any] | None = None
+
+
 class LitellmModel(Model):
     """This class enables using any model via LiteLLM. LiteLLM allows you to acess OpenAPI,
     Anthropic, Gemini, Mistral, and many other models.
@@ -168,9 +177,15 @@ async def get_response(
                 "output_tokens": usage.output_tokens,
             }
 
+            # Build provider_data for provider specific fields
+            provider_data: dict[str, Any] = {"model": self.model}
+            if message is not None and hasattr(response, "id"):
+                provider_data["response_id"] = response.id
+
             items = (
                 Converter.message_to_output_items(
-                    LitellmConverter.convert_message_to_openai(message)
+                    LitellmConverter.convert_message_to_openai(message, model=self.model),
+                    provider_data=provider_data,
                 )
                 if message is not None
                 else []
@@ -215,7 +230,9 @@ async def stream_response(
             )
 
             final_response: Response | None = None
-            async for chunk in ChatCmplStreamHandler.handle_stream(response, stream):
+            async for chunk in ChatCmplStreamHandler.handle_stream(
+                response, stream, model=self.model
+            ):
                 yield chunk
 
                 if chunk.type == "response.completed":
@@ -283,12 +300,19 @@ async def _fetch_response(
             input,
             preserve_thinking_blocks=preserve_thinking_blocks,
             preserve_tool_output_all_content=True,
+            model=self.model,
         )
 
         # Fix for interleaved thinking bug: reorder messages to ensure tool_use comes before tool_result  # noqa: E501
         if "anthropic" in self.model.lower() or "claude" in self.model.lower():
             converted_messages = self._fix_tool_message_ordering(converted_messages)
 
+        # Convert Google's extra_content to litellm's provider_specific_fields format
+        if "gemini" in self.model.lower():
+            converted_messages = self._convert_gemini_extra_content_to_provider_specific_fields(
+                converted_messages
+            )
+
         if system_instructions:
             converted_messages.insert(
                 0,
@@ -438,6 +462,65 @@ async def _fetch_response(
         )
         return response, ret
 
+    def _convert_gemini_extra_content_to_provider_specific_fields(
+        self, messages: list[ChatCompletionMessageParam]
+    ) -> list[ChatCompletionMessageParam]:
+        """
+        Convert Gemini model's extra_content format to provider_specific_fields format for litellm.
+
+        Transforms tool calls from internal format:
+            extra_content={"google": {"thought_signature": "..."}}
+        To litellm format:
+            provider_specific_fields={"thought_signature": "..."}
+
+        Only processes tool_calls that appear after the last user message.
+        See: https://ai.google.dev/gemini-api/docs/thought-signatures
+        """
+
+        # Find the index of the last user message
+        last_user_index = -1
+        for i in range(len(messages) - 1, -1, -1):
+            if isinstance(messages[i], dict) and messages[i].get("role") == "user":
+                last_user_index = i
+                break
+
+        for i, message in enumerate(messages):
+            if not isinstance(message, dict):
+                continue
+
+            # Only process assistant messages that come after the last user message
+            # If no user message found (last_user_index == -1), process all messages
+            if last_user_index != -1 and i <= last_user_index:
+                continue
+
+            # Check if this is an assistant message with tool calls
+            if message.get("role") == "assistant" and message.get("tool_calls"):
+                tool_calls = message.get("tool_calls", [])
+
+                for tool_call in tool_calls:  # type: ignore[attr-defined]
+                    if not isinstance(tool_call, dict):
+                        continue
+
+                    # Default to skip validator, overridden if valid thought signature exists
+                    tool_call["provider_specific_fields"] = {
+                        "thought_signature": "skip_thought_signature_validator"
+                    }
+
+                    # Override with actual thought signature if extra_content exists
+                    if "extra_content" in tool_call:
+                        extra_content = tool_call.pop("extra_content")
+                        if isinstance(extra_content, dict):
+                            # Extract google-specific fields
+                            google_fields = extra_content.get("google")
+                            if google_fields and isinstance(google_fields, dict):
+                                thought_sig = google_fields.get("thought_signature")
+                                if thought_sig:
+                                    tool_call["provider_specific_fields"] = {
+                                        "thought_signature": thought_sig
+                                    }
+
+        return messages
+
     def _fix_tool_message_ordering(
         self, messages: list[ChatCompletionMessageParam]
     ) -> list[ChatCompletionMessageParam]:
@@ -565,15 +648,26 @@ def _merge_headers(self, model_settings: ModelSettings):
 class LitellmConverter:
     @classmethod
     def convert_message_to_openai(
-        cls, message: litellm.types.utils.Message
+        cls, message: litellm.types.utils.Message, model: str | None = None
     ) -> ChatCompletionMessage:
+        """
+        Convert a LiteLLM message to OpenAI ChatCompletionMessage format.
+
+        Args:
+            message: The LiteLLM message to convert
+            model: The target model to convert to. Used to handle provider-specific
+                transformations.
+        """
         if message.role != "assistant":
             raise ModelBehaviorError(f"Unsupported role: {message.role}")
 
         tool_calls: (
             list[ChatCompletionMessageFunctionToolCall | ChatCompletionMessageCustomToolCall] | None
         ) = (
-            [LitellmConverter.convert_tool_call_to_openai(tool) for tool in message.tool_calls]
+            [
+                LitellmConverter.convert_tool_call_to_openai(tool, model=model)
+                for tool in message.tool_calls
+            ]
             if message.tool_calls
             else None
         )
@@ -643,13 +737,43 @@ def convert_annotations_to_openai(
 
     @classmethod
     def convert_tool_call_to_openai(
-        cls, tool_call: litellm.types.utils.ChatCompletionMessageToolCall
+        cls, tool_call: litellm.types.utils.ChatCompletionMessageToolCall, model: str | None = None
     ) -> ChatCompletionMessageFunctionToolCall:
-        return ChatCompletionMessageFunctionToolCall(
-            id=tool_call.id,
+        # Clean up litellm's addition of __thought__ suffix to tool_call.id for
+        # Gemini models. See: https://github.com/BerriAI/litellm/pull/16895
+        # This suffix is redundant since we can get thought_signature from
+        # provider_specific_fields, and this hack causes validation errors when
+        # cross-model passing to other models.
+        tool_call_id = tool_call.id
+        if model and "gemini" in model.lower() and "__thought__" in tool_call_id:
+            tool_call_id = tool_call_id.split("__thought__")[0]
+
+        # Convert litellm's tool call format to chat completion message format
+        base_tool_call = ChatCompletionMessageFunctionToolCall(
+            id=tool_call_id,
             type="function",
             function=Function(
                 name=tool_call.function.name or "",
                 arguments=tool_call.function.arguments,
             ),
         )
+
+        # Preserve provider-specific fields if present (e.g., Gemini thought signatures)
+        if hasattr(tool_call, "provider_specific_fields") and tool_call.provider_specific_fields:
+            # Convert to nested extra_content structure
+            extra_content: dict[str, Any] = {}
+            provider_fields = tool_call.provider_specific_fields
+
+            # Check for thought_signature (Gemini specific)
+            if model and "gemini" in model.lower():
+                if "thought_signature" in provider_fields:
+                    extra_content["google"] = {
+                        "thought_signature": provider_fields["thought_signature"]
+                    }
+
+            return InternalToolCall(
+                **base_tool_call.model_dump(),
+                extra_content=extra_content if extra_content else None,
+            )
+
+        return base_tool_call

src/agents/handoffs/history.py

@@ -144,7 +144,7 @@ def _format_transcript_item(item: TResponseInputItem) -> str:
         return f"{prefix}: {content_str}" if content_str else prefix
 
     item_type = item.get("type", "item")
-    rest = {k: v for k, v in item.items() if k != "type"}
+    rest = {k: v for k, v in item.items() if k not in ("type", "provider_data")}
     try:
         serialized = json.dumps(rest, ensure_ascii=False, default=str)
     except TypeError: