Skip to content

The agent engineering platform.

fix(fireworks): `bind_tools(strict: bool)` and `reasoning_content` (#34343)

Dragos Bobolea6d447f89d9cc
diff --git a/libs/partners/fireworks/langchain_fireworks/chat_models.py b/libs/partners/fireworks/langchain_fireworks/chat_models.py
index ec8d65f081..37c8c4bc48 100644
--- a/libs/partners/fireworks/langchain_fireworks/chat_models.py
+++ b/libs/partners/fireworks/langchain_fireworks/chat_models.py
@@ -114,8 +114,12 @@ def _convert_dict_to_message(_dict: Mapping[str, Any]) -> BaseMessage:
# Also Fireworks returns None for tool invocations
content = _dict.get("content", "") or ""
additional_kwargs: dict = {}
+ if reasoning_content := _dict.get("reasoning_content"):
+ additional_kwargs["reasoning_content"] = reasoning_content
+
if function_call := _dict.get("function_call"):
additional_kwargs["function_call"] = dict(function_call)
+
tool_calls = []
invalid_tool_calls = []
if raw_tool_calls := _dict.get("tool_calls"):
@@ -678,7 +682,10 @@ class ChatFireworks(BaseChatModel):
**kwargs: Any additional parameters to pass to
`langchain_fireworks.chat_models.ChatFireworks.bind`
""" # noqa: E501
- formatted_tools = [convert_to_openai_tool(tool) for tool in tools]
+ strict = kwargs.pop("strict", None)
+ formatted_tools = [
+ convert_to_openai_tool(tool, strict=strict) for tool in tools
+ ]
if tool_choice is not None and tool_choice:
if isinstance(tool_choice, str) and (
tool_choice not in ("auto", "any", "none")
diff --git a/libs/partners/fireworks/tests/integration_tests/test_chat_models.py b/libs/partners/fireworks/tests/integration_tests/test_chat_models.py
index 2cd6539388..112fe77675 100644
--- a/libs/partners/fireworks/tests/integration_tests/test_chat_models.py
+++ b/libs/partners/fireworks/tests/integration_tests/test_chat_models.py
@@ -18,15 +18,30 @@ from langchain_fireworks import ChatFireworks
_MODEL = "accounts/fireworks/models/gpt-oss-120b"
-def test_tool_choice_bool() -> None:
- """Test that tool choice is respected just passing in True."""
[email protected]("strict", [None, True, False])
+def test_tool_choice_bool(strict: bool | None) -> None: # noqa: FBT001
+ """Test that tool choice is respected with different strict values."""
llm = ChatFireworks(model="fireworks/kimi-k2-instruct-0905")
class MyTool(BaseModel):
name: str
age: int
- with_tool = llm.bind_tools([MyTool], tool_choice=True)
+ kwargs = {"tool_choice": True}
+ if strict is not None:
+ kwargs["strict"] = strict
+ with_tool = llm.bind_tools([MyTool], **kwargs)
+
+ # Verify that strict is correctly set in the tool definition
+ assert hasattr(with_tool, "kwargs")
+ tools = with_tool.kwargs.get("tools", [])
+ assert len(tools) == 1
+ tool_def = tools[0]
+ assert "function" in tool_def
+ if strict is None:
+ assert "strict" not in tool_def["function"]
+ else:
+ assert tool_def["function"].get("strict") is strict
resp = with_tool.invoke("Who was the 27 year old named Erick?")
assert isinstance(resp, AIMessage)
diff --git a/libs/partners/fireworks/tests/unit_tests/test_chat_models.py b/libs/partners/fireworks/tests/unit_tests/test_chat_models.py
new file mode 100644
index 0000000000..948b5e45c5
--- /dev/null
+++ b/libs/partners/fireworks/tests/unit_tests/test_chat_models.py
@@ -0,0 +1,38 @@
+"""Unit tests for ChatFireworks."""
+
+from __future__ import annotations
+
+from langchain_core.messages import AIMessage
+
+from langchain_fireworks.chat_models import _convert_dict_to_message
+
+
+def test_convert_dict_to_message_with_reasoning_content() -> None:
+ """Test that reasoning_content is correctly extracted from API response."""
+ response_dict = {
+ "role": "assistant",
+ "content": "The answer is 42.",
+ "reasoning_content": "Let me think about this step by step...",
+ }
+
+ message = _convert_dict_to_message(response_dict)
+
+ assert isinstance(message, AIMessage)
+ assert message.content == "The answer is 42."
+ assert "reasoning_content" in message.additional_kwargs
+ expected_reasoning = "Let me think about this step by step..."
+ assert message.additional_kwargs["reasoning_content"] == expected_reasoning
+
+
+def test_convert_dict_to_message_without_reasoning_content() -> None:
+ """Test that messages without reasoning_content work correctly."""
+ response_dict = {
+ "role": "assistant",
+ "content": "The answer is 42.",
+ }
+
+ message = _convert_dict_to_message(response_dict)
+
+ assert isinstance(message, AIMessage)
+ assert message.content == "The answer is 42."
+ assert "reasoning_content" not in message.additional_kwargs