Chat Models

Chat model initialization and configuration

init_chat_model

def init_chat_model(
    model: str | None = None,
    *,
    model_provider: str | None = None,
    configurable_fields: Literal["any"] | list[str] | tuple[str, ...] | None = None,
    config_prefix: str | None = None,
    **kwargs: Any,
) -> BaseChatModel | _ConfigurableModel

Initialize a chat model from any supported provider using a unified interface.

Requires the integration package for the chosen model provider to be installed (e.g., pip install langchain-openai).

Refer to the provider integration’s API reference for supported model parameters to use as **kwargs.

Two main use cases

1. Fixed model – specify the model upfront and get a ready-to-use chat model.
2. Configurable model – choose to specify parameters (including model name) at runtime via config. Makes it easy to switch between models/providers without changing your code.

model · str | None

The model name, optionally prefixed with provider (e.g., 'openai:gpt-4o').

Prefer exact model IDs from provider docs over aliases for reliable behavior (e.g., dated versions like '...-20250514' instead of '...-latest').

Will attempt to infer model_provider from model if not specified.

The following providers will be inferred based on these model prefixes:

• gpt-... | o1... | o3... → openai
• claude... → anthropic
• amazon... → bedrock
• gemini... → google_vertexai
• command... → cohere
• accounts/fireworks... → fireworks
• mistral... → mistralai
• deepseek... → deepseek
• grok... → xai
• sonar... → perplexity
• solar... → upstage

model_provider · str | None

The model provider if not specified as part of the model arg (see above).

Supported model_provider values and the corresponding integration package are:

• openai → langchain-openai
• anthropic → langchain-anthropic
• azure_openai → langchain-openai
• azure_ai → langchain-azure-ai
• google_vertexai → langchain-google-vertexai
• google_genai → langchain-google-genai
• anthropic_bedrock → langchain-aws
• bedrock → langchain-aws
• bedrock_converse → langchain-aws
• cohere → langchain-cohere
• fireworks → langchain-fireworks
• together → langchain-together
• mistralai → langchain-mistralai
• huggingface → langchain-huggingface
• groq → langchain-groq
• ollama → langchain-ollama
• google_anthropic_vertex → langchain-google-vertexai
• deepseek → langchain-deepseek
• ibm → langchain-ibm
• nvidia → langchain-nvidia-ai-endpoints
• xai → langchain-xai
• openrouter → langchain-openrouter
• perplexity → langchain-perplexity
• upstage → langchain-upstage

configurable_fields · Literal['any'] | list[str] | tuple[str, ...] | None

Which model parameters are configurable at runtime:

• None: No configurable fields (i.e., a fixed model).
• 'any': All fields are configurable. See security note below.
• list[str] | Tuple[str, ...]: Specified fields are configurable.

Fields are assumed to have config_prefix stripped if a config_prefix is specified.

If model is specified, then defaults to None.

If model is not specified, then defaults to ("model", "model_provider").

Setting configurable_fields="any" means fields like api_key, base_url, etc., can be altered at runtime, potentially redirecting model requests to a different service/user.

Make sure that if you’re accepting untrusted configurations that you enumerate the configurable_fields=(...) explicitly.

config_prefix · str | None

Optional prefix for configuration keys.

Useful when you have multiple configurable models in the same application.

If 'config_prefix' is a non-empty string then model will be configurable at runtime via the config["configurable"]["{config_prefix}_{param}"] keys. See examples below.

If 'config_prefix' is an empty string then model will be configurable via config["configurable"]["{param}"].

**kwargs · Any

Additional model-specific keyword args to pass to the underlying chat model’s __init__ method. Common parameters include:

• temperature: Model temperature for controlling randomness.
• max_tokens: Maximum number of output tokens.
• timeout: Maximum time (in seconds) to wait for a response.
• max_retries: Maximum number of retry attempts for failed requests.
• base_url: Custom API endpoint URL.
• rate_limiter: A BaseRateLimiter instance to control request rate.

Refer to the specific model provider’s integration reference for all available parameters.

returns · BaseChatModel | _ConfigurableModel

A BaseChatModel corresponding to the model_name and model_provider specified if configurability is inferred to be False. If configurable, a chat model emulator that initializes the underlying model at runtime once a config is passed in.

Examples

Initialize a non-configurable model

# pip install langchain langchain-openai langchain-anthropic langchain-google-vertexai

from langchain.chat_models import init_chat_model

o3_mini = init_chat_model("openai:o3-mini", temperature=0)
claude_sonnet = init_chat_model("anthropic:claude-sonnet-4-5-20250929", temperature=0)
gemini_2_5_flash = init_chat_model("google_vertexai:gemini-2.5-flash", temperature=0)

o3_mini.invoke("what's your name")
claude_sonnet.invoke("what's your name")
gemini_2_5_flash.invoke("what's your name")

Partially configurable model with no default

# pip install langchain langchain-openai langchain-anthropic

from langchain.chat_models import init_chat_model

# (We don't need to specify configurable=True if a model isn't specified.)
configurable_model = init_chat_model(temperature=0)

configurable_model.invoke("what's your name", config={"configurable": {"model": "gpt-4o"}})
# Use GPT-4o to generate the response

configurable_model.invoke(
    "what's your name",
    config={"configurable": {"model": "claude-sonnet-4-5-20250929"}},
)

Fully configurable model with a default

# pip install langchain langchain-openai langchain-anthropic

from langchain.chat_models import init_chat_model

configurable_model_with_default = init_chat_model(
    "openai:gpt-4o",
    configurable_fields="any",  # This allows us to configure other params like temperature, max_tokens, etc at runtime.
    config_prefix="foo",
    temperature=0,
)

configurable_model_with_default.invoke("what's your name")
# GPT-4o response with temperature 0 (as set in default)

configurable_model_with_default.invoke(
    "what's your name",
    config={
        "configurable": {
            "foo_model": "anthropic:claude-sonnet-4-5-20250929",
            "foo_temperature": 0.6,
        }
    },
)
# Override default to use Sonnet 4.5 with temperature 0.6 to generate response

Bind tools to a configurable model

You can call any chat model declarative methods on a configurable model in the same way that you would with a normal model:

# pip install langchain langchain-openai langchain-anthropic

from langchain.chat_models import init_chat_model
from pydantic import BaseModel, Field


class GetWeather(BaseModel):
    '''Get the current weather in a given location'''

    location: str = Field(..., description="The city and state, e.g. San Francisco, CA")


class GetPopulation(BaseModel):
    '''Get the current population in a given location'''

    location: str = Field(..., description="The city and state, e.g. San Francisco, CA")


configurable_model = init_chat_model(
    "gpt-4o", configurable_fields=("model", "model_provider"), temperature=0
)

configurable_model_with_tools = configurable_model.bind_tools(
    [
        GetWeather,
        GetPopulation,
    ]
)
configurable_model_with_tools.invoke(
    "Which city is hotter today and which is bigger: LA or NY?"
)
# Use GPT-4o

configurable_model_with_tools.invoke(
    "Which city is hotter today and which is bigger: LA or NY?",
    config={"configurable": {"model": "claude-sonnet-4-5-20250929"}},
)
# Use Sonnet 4.5

BaseChatModel

from langchain_core.language_models import BaseChatModel

The base class for all chat models in LangChain. This is re-exported from langchain_core for convenience.

See the LangChain Core documentation for the full API reference.