Implementation:Langchain ai Langchain BaseChatOpenAI Bind Tools

Knowledge Sources	LangChain
Domains	NLP, Tool_Use
Last Updated	2026-02-11 00:00 GMT

Overview

Concrete tool for binding tool schemas to OpenAI-compatible chat models provided by the LangChain OpenAI integration.

Description

The BaseChatOpenAI.bind_tools() method accepts a sequence of tool definitions (functions, Pydantic models, BaseTool instances, or raw dicts), converts each to OpenAI's tool schema format via convert_to_openai_tool(), and returns a new Runnable with tools bound as kwargs. The convert_to_openai_tool() utility handles the schema conversion, including optional strict mode for guaranteed schema compliance.

Usage

Call bind_tools() on any OpenAI-compatible chat model to enable tool calling. For Anthropic models, ChatAnthropic.bind_tools() follows the same interface but converts to Anthropic's tool format.

Code Reference

Source Location

Repository: langchain
File: libs/partners/openai/langchain_openai/chat_models/base.py (bind_tools), libs/core/langchain_core/utils/function_calling.py (convert_to_openai_tool)
Lines: base.py L1867-1959 (bind_tools); function_calling.py L498-575 (convert_to_openai_tool)

Signature

def bind_tools(
    self,
    tools: Sequence[dict[str, Any] | type | Callable | BaseTool],
    *,
    tool_choice: dict | str | bool | None = None,
    strict: bool | None = None,
    parallel_tool_calls: bool | None = None,
    response_format: _DictOrPydanticClass | None = None,
    **kwargs: Any,
) -> Runnable[LanguageModelInput, AIMessage]:

def convert_to_openai_tool(
    tool: Mapping[str, Any] | type[BaseModel] | Callable | BaseTool,
    *,
    strict: bool | None = None,
) -> dict[str, Any]:

Import

from langchain_openai import ChatOpenAI
# convert_to_openai_tool is used internally; import if needed:
from langchain_core.utils.function_calling import convert_to_openai_tool

I/O Contract

Inputs

Name	Type	Required	Description
tools	Sequence[dict or type or Callable or BaseTool]	Yes	Tool definitions to bind
tool_choice	dict or str or bool or None	No	Force tool usage ("auto", "any", "required", or tool name)
strict	bool or None	No	Enforce strict JSON schema compliance
parallel_tool_calls	bool or None	No	Allow parallel tool calls in one turn

Outputs

Name	Type	Description
return	Runnable[LanguageModelInput, AIMessage]	Model with tools bound, ready for invocation

Usage Examples

Binding Tools

from langchain_openai import ChatOpenAI
from langchain_core.tools import tool

@tool
def get_weather(city: str) -> str:
    """Get weather for a city."""
    return f"Sunny in {city}"

llm = ChatOpenAI(model="gpt-4o-mini")
llm_with_tools = llm.bind_tools([get_weather])

response = llm_with_tools.invoke("What's the weather in Paris?")
print(response.tool_calls)
# [{'name': 'get_weather', 'args': {'city': 'Paris'}, 'id': 'call_...'}]

Forcing Tool Choice

# Force the model to use a specific tool
llm_forced = llm.bind_tools(
    [get_weather],
    tool_choice="get_weather",
)

# Force the model to use any tool (must use at least one)
llm_any = llm.bind_tools(
    [get_weather],
    tool_choice="required",
)

Related Pages

Implements Principle

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment