AzureChatOpenAI#

class langchain_openai.chat_models.azure.AzureChatOpenAI[source]#

Bases: BaseChatOpenAI

Azure OpenAI chat model integration.

Setup:

Head to the https://learn.microsoft.com/en-us/azure/ai-services/openai/chatgpt-quickstart?tabs=command-line%2Cpython-new&pivots=programming-language-python to create your Azure OpenAI deployment.

Then install langchain-openai and set environment variables AZURE_OPENAI_API_KEY and AZURE_OPENAI_ENDPOINT:

pip install -U langchain-openai

export AZURE_OPENAI_API_KEY="your-api-key"
export AZURE_OPENAI_ENDPOINT="https://your-endpoint.openai.azure.com/"

Key init args — completion params:

azure_deployment: str: Name of Azure OpenAI deployment to use.
temperature: float: Sampling temperature.
max_tokens: Optional[int]: Max number of tokens to generate.
logprobs: Optional[bool]: Whether to return logprobs.

Key init args — client params:

api_version: str: Azure OpenAI API version to use. See more on the different versions here: https://learn.microsoft.com/en-us/azure/ai-services/openai/reference#rest-api-versioning
timeout: Union[float, Tuple[float, float], Any, None]: Timeout for requests.
max_retries: int: Max number of retries.
organization: Optional[str]: OpenAI organization ID. If not passed in will be read from env var OPENAI_ORG_ID.
model: Optional[str]: The name of the underlying OpenAI model. Used for tracing and token counting. Does not affect completion. E.g. “gpt-4”, “gpt-35-turbo”, etc.
model_version: Optional[str]: The version of the underlying OpenAI model. Used for tracing and token counting. Does not affect completion. E.g., “0125”, “0125-preview”, etc.

See full list of supported init args and their descriptions in the params section.

Instantiate:

from langchain_openai import AzureChatOpenAI

llm = AzureChatOpenAI(
    azure_deployment="your-deployment",
    api_version="2024-05-01-preview",
    temperature=0,
    max_tokens=None,
    timeout=None,
    max_retries=2,
    # organization="...",
    # model="gpt-35-turbo",
    # model_version="0125",
    # other params...
)

NOTE: Any param which is not explicitly supported will be passed directly to the openai.AzureOpenAI.chat.completions.create(...) API every time to the model is invoked. For example:

from langchain_openai import AzureChatOpenAI
import openai

AzureChatOpenAI(..., logprobs=True).invoke(...)

# results in underlying API call of:

openai.AzureOpenAI(..).chat.completions.create(..., logprobs=True)

# which is also equivalent to:

AzureChatOpenAI(...).invoke(..., logprobs=True)

Invoke:

messages = [
    (
        "system",
        "You are a helpful translator. Translate the user sentence to French.",
    ),
    ("human", "I love programming."),
]
llm.invoke(messages)

AIMessage(
    content="J'adore programmer.",
    usage_metadata={"input_tokens": 28, "output_tokens": 6, "total_tokens": 34},
    response_metadata={
        "token_usage": {
            "completion_tokens": 6,
            "prompt_tokens": 28,
            "total_tokens": 34,
        },
        "model_name": "gpt-4",
        "system_fingerprint": "fp_7ec89fabc6",
        "prompt_filter_results": [
            {
                "prompt_index": 0,
                "content_filter_results": {
                    "hate": {"filtered": False, "severity": "safe"},
                    "self_harm": {"filtered": False, "severity": "safe"},
                    "sexual": {"filtered": False, "severity": "safe"},
                    "violence": {"filtered": False, "severity": "safe"},
                },
            }
        ],
        "finish_reason": "stop",
        "logprobs": None,
        "content_filter_results": {
            "hate": {"filtered": False, "severity": "safe"},
            "self_harm": {"filtered": False, "severity": "safe"},
            "sexual": {"filtered": False, "severity": "safe"},
            "violence": {"filtered": False, "severity": "safe"},
        },
    },
    id="run-6d7a5282-0de0-4f27-9cc0-82a9db9a3ce9-0",
)

Stream:

for chunk in llm.stream(messages):
    print(chunk)

AIMessageChunk(content="", id="run-a6f294d3-0700-4f6a-abc2-c6ef1178c37f")
AIMessageChunk(content="J", id="run-a6f294d3-0700-4f6a-abc2-c6ef1178c37f")
AIMessageChunk(content="'", id="run-a6f294d3-0700-4f6a-abc2-c6ef1178c37f")
AIMessageChunk(content="ad", id="run-a6f294d3-0700-4f6a-abc2-c6ef1178c37f")
AIMessageChunk(content="ore", id="run-a6f294d3-0700-4f6a-abc2-c6ef1178c37f")
AIMessageChunk(content=" la", id="run-a6f294d3-0700-4f6a-abc2-c6ef1178c37f")
AIMessageChunk(content=" programm", id="run-a6f294d3-0700-4f6a-abc2-c6ef1178c37f")
AIMessageChunk(content="ation", id="run-a6f294d3-0700-4f6a-abc2-c6ef1178c37f")
AIMessageChunk(content=".", id="run-a6f294d3-0700-4f6a-abc2-c6ef1178c37f")
AIMessageChunk(
    content="",
    response_metadata={
        "finish_reason": "stop",
        "model_name": "gpt-4",
        "system_fingerprint": "fp_811936bd4f",
    },
    id="run-a6f294d3-0700-4f6a-abc2-c6ef1178c37f",
)

stream = llm.stream(messages)
full = next(stream)
for chunk in stream:
    full += chunk
full

AIMessageChunk(
    content="J'adore la programmation.",
    response_metadata={
        "finish_reason": "stop",
        "model_name": "gpt-4",
        "system_fingerprint": "fp_811936bd4f",
    },
    id="run-ba60e41c-9258-44b8-8f3a-2f10599643b3",
)

Async:

await llm.ainvoke(messages)

# stream:
# async for chunk in (await llm.astream(messages))

# batch:
# await llm.abatch([messages])

Tool calling:

from langchain_core.pydantic_v1 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"
    )


llm_with_tools = llm.bind_tools([GetWeather, GetPopulation])
ai_msg = llm_with_tools.invoke(
    "Which city is hotter today and which is bigger: LA or NY?"
)
ai_msg.tool_calls

[
    {
        "name": "GetWeather",
        "args": {"location": "Los Angeles, CA"},
        "id": "call_6XswGD5Pqk8Tt5atYr7tfenU",
    },
    {
        "name": "GetWeather",
        "args": {"location": "New York, NY"},
        "id": "call_ZVL15vA8Y7kXqOy3dtmQgeCi",
    },
    {
        "name": "GetPopulation",
        "args": {"location": "Los Angeles, CA"},
        "id": "call_49CFW8zqC9W7mh7hbMLSIrXw",
    },
    {
        "name": "GetPopulation",
        "args": {"location": "New York, NY"},
        "id": "call_6ghfKxV264jEfe1mRIkS3PE7",
    },
]

Structured output:

from typing import Optional

from langchain_core.pydantic_v1 import BaseModel, Field


class Joke(BaseModel):
    '''Joke to tell user.'''

    setup: str = Field(description="The setup of the joke")
    punchline: str = Field(description="The punchline to the joke")
    rating: Optional[int] = Field(description="How funny the joke is, from 1 to 10")


structured_llm = llm.with_structured_output(Joke)
structured_llm.invoke("Tell me a joke about cats")

Joke(
    setup="Why was the cat sitting on the computer?",
    punchline="To keep an eye on the mouse!",
    rating=None,
)

See AzureChatOpenAI.with_structured_output() for more.

JSON mode:

json_llm = llm.bind(response_format={"type": "json_object"})
ai_msg = json_llm.invoke(
    "Return a JSON object with key 'random_ints' and a value of 10 random ints in [0-99]"
)
ai_msg.content

'\n{\n  "random_ints": [23, 87, 45, 12, 78, 34, 56, 90, 11, 67]\n}'

Image input:

import base64
import httpx
from langchain_core.messages import HumanMessage

image_url = "https://upload.wikimedia.org/wikipedia/commons/thumb/d/dd/Gfp-wisconsin-madison-the-nature-boardwalk.jpg/2560px-Gfp-wisconsin-madison-the-nature-boardwalk.jpg"
image_data = base64.b64encode(httpx.get(image_url).content).decode("utf-8")
message = HumanMessage(
    content=[
        {"type": "text", "text": "describe the weather in this image"},
        {
            "type": "image_url",
            "image_url": {"url": f"data:image/jpeg;base64,{image_data}"},
        },
    ]
)
ai_msg = llm.invoke([message])
ai_msg.content

"The weather in the image appears to be quite pleasant. The sky is mostly clear"

Token usage:

ai_msg = llm.invoke(messages)
ai_msg.usage_metadata

{"input_tokens": 28, "output_tokens": 5, "total_tokens": 33}

Logprobs:

logprobs_llm = llm.bind(logprobs=True)
ai_msg = logprobs_llm.invoke(messages)
ai_msg.response_metadata["logprobs"]

{
    "content": [
        {
            "token": "J",
            "bytes": [74],
            "logprob": -4.9617593e-06,
            "top_logprobs": [],
        },
        {
            "token": "'adore",
            "bytes": [39, 97, 100, 111, 114, 101],
            "logprob": -0.25202933,
            "top_logprobs": [],
        },
        {
            "token": " la",
            "bytes": [32, 108, 97],
            "logprob": -0.20141791,
            "top_logprobs": [],
        },
        {
            "token": " programmation",
            "bytes": [
                32,
                112,
                114,
                111,
                103,
                114,
                97,
                109,
                109,
                97,
                116,
                105,
                111,
                110,
            ],
            "logprob": -1.9361265e-07,
            "top_logprobs": [],
        },
        {
            "token": ".",
            "bytes": [46],
            "logprob": -1.2233183e-05,
            "top_logprobs": [],
        },
    ]
}

Response metadata

ai_msg = llm.invoke(messages)
ai_msg.response_metadata

{
    "token_usage": {
        "completion_tokens": 6,
        "prompt_tokens": 28,
        "total_tokens": 34,
    },
    "model_name": "gpt-35-turbo",
    "system_fingerprint": None,
    "prompt_filter_results": [
        {
            "prompt_index": 0,
            "content_filter_results": {
                "hate": {"filtered": False, "severity": "safe"},
                "self_harm": {"filtered": False, "severity": "safe"},
                "sexual": {"filtered": False, "severity": "safe"},
                "violence": {"filtered": False, "severity": "safe"},
            },
        }
    ],
    "finish_reason": "stop",
    "logprobs": None,
    "content_filter_results": {
        "hate": {"filtered": False, "severity": "safe"},
        "self_harm": {"filtered": False, "severity": "safe"},
        "sexual": {"filtered": False, "severity": "safe"},
        "violence": {"filtered": False, "severity": "safe"},
    },
}

Note

AzureChatOpenAI implements the standard Runnable Interface. 🏃

The Runnable Interface has additional methods that are available on runnables, such as with_types, with_retry, assign, bind, get_graph, and more.

param azure_ad_token: SecretStr | None [Optional]#

Your Azure Active Directory token.

Automatically inferred from env var AZURE_OPENAI_AD_TOKEN if not provided.

For more: https://www.microsoft.com/en-us/security/business/identity-access/microsoft-entra-id.

Constraints:

type = string
writeOnly = True
format = password

param azure_ad_token_provider: Callable[[], str] | None = None#

A function that returns an Azure Active Directory token.

Will be invoked on every request.

param azure_endpoint: str | None [Optional]#

Your Azure endpoint, including the resource.

Automatically inferred from env var AZURE_OPENAI_ENDPOINT if not provided.

Example: https://example-resource.azure.openai.com/

param cache: BaseCache | bool | None = None#

Whether to cache the response.

If true, will use the global cache.
If false, will not use a cache
If None, will use the global cache if it’s set, otherwise no cache.
If instance of BaseCache, will use the provided cache.

Caching is not currently supported for streaming methods of models.

param callback_manager: BaseCallbackManager | None = None#: Deprecated since version 0.1.7: Use callbacks instead.

Callback manager to add to the run trace.

param callbacks: Callbacks = None#: Callbacks to add to the run trace.

param custom_get_token_ids: Callable[[str], List[int]] | None = None#: Optional encoder to use for counting tokens.

param default_headers: Mapping[str, str] | None = None#

param default_query: Mapping[str, object] | None = None#

param deployment_name: str | None = None (alias 'azure_deployment')#

A model deployment.

If given sets the base client URL to include /deployments/{azure_deployment}. Note: this means you won’t be able to use non-deployment endpoints.

param disable_streaming: bool | Literal['tool_calling'] = False#

Whether to disable streaming for this model.

If streaming is bypassed, then stream()/astream() will defer to invoke()/ainvoke().

If True, will always bypass streaming case.
If “tool_calling”, will bypass streaming case only when the model is called with a tools keyword argument.
If False (default), will always use streaming case if available.

param extra_body: Mapping[str, Any] | None = None#: Optional additional JSON properties to include in the request parameters when making requests to OpenAI compatible APIs, such as vLLM.

param frequency_penalty: float | None = None#: Penalizes repeated tokens according to frequency.

param http_async_client: Any | None = None#: Optional httpx.AsyncClient. Only used for async invocations. Must specify http_client as well if you’d like a custom client for sync invocations.

param http_client: Any | None = None#: Optional httpx.Client. Only used for sync invocations. Must specify http_async_client as well if you’d like a custom client for async invocations.

param include_response_headers: bool = False#: Whether to include response headers in the output message response_metadata.

param logit_bias: Dict[int, int] | None = None#: Modify the likelihood of specified tokens appearing in the completion.

param logprobs: bool | None = None#: Whether to return logprobs.

param max_retries: int = 2#: Maximum number of retries to make when generating.

param max_tokens: int | None = None#: Maximum number of tokens to generate.

param metadata: Dict[str, Any] | None = None#: Metadata to add to the run trace.

param model_kwargs: Dict[str, Any] [Optional]#: Holds any model parameters valid for create call not explicitly specified.

param model_name: str | None = None (alias 'model')#

Name of the deployed OpenAI model, e.g. “gpt-4o”, “gpt-35-turbo”, etc.

Distinct from the Azure deployment name, which is set by the Azure user. Used for tracing and token counting. Does NOT affect completion.

param model_version: str = ''#

The version of the model (e.g. “0125” for gpt-3.5-0125).

Azure OpenAI doesn’t return model version with the response by default so it must be manually specified if you want to use this information downstream, e.g. when calculating costs.

When you specify the version, it will be appended to the model name in the response. Setting correct version will help you to calculate the cost properly. Model version is not validated, so make sure you set it correctly to get the correct cost.

param n: int = 1#: Number of chat completions to generate for each prompt.

param openai_api_base: str | None = None (alias 'base_url')#: Base URL path for API requests, leave blank if not using a proxy or service emulator.

param openai_api_key: SecretStr | None [Optional] (alias 'api_key')#

Automatically inferred from env var AZURE_OPENAI_API_KEY if not provided.

Constraints:

type = string
writeOnly = True
format = password

param openai_api_type: str | None [Optional]#: Legacy, for openai<1.0.0 support.

param openai_api_version: str | None [Optional] (alias 'api_version')#: Automatically inferred from env var OPENAI_API_VERSION if not provided.

param openai_organization: str | None = None (alias 'organization')#: Automatically inferred from env var OPENAI_ORG_ID if not provided.

param openai_proxy: str | None = None#

param presence_penalty: float | None = None#: Penalizes repeated tokens.

param rate_limiter: BaseRateLimiter | None = None#: An optional rate limiter to use for limiting the number of requests.

param request_timeout: float | Tuple[float, float] | Any | None = None (alias 'timeout')#: Timeout for requests to OpenAI completion API. Can be float, httpx.Timeout or None.

param seed: int | None = None#: Seed for generation

param stop: List[str] | str | None = None (alias 'stop_sequences')#: Default stop sequences.

param streaming: bool = False#: Whether to stream the results or not.

param tags: List[str] | None = None#: Tags to add to the run trace.

param temperature: float = 0.7#: What sampling temperature to use.

param tiktoken_model_name: str | None = None#: The model name to pass to tiktoken when using this class. Tiktoken is used to count the number of tokens in documents to constrain them to be under a certain limit. By default, when set to None, this will be the same as the embedding model name. However, there are some cases where you may want to use this Embedding class with a model name not supported by tiktoken. This can include when using Azure embeddings or when using one of the many model providers that expose an OpenAI-like API but with different models. In those cases, in order to avoid erroring when tiktoken is called, you can specify a model name to use here.

param top_logprobs: int | None = None#: Number of most likely tokens to return at each token position, each with an associated log probability. logprobs must be set to true if this parameter is used.

param top_p: float | None = None#: Total probability mass of tokens to consider at each step.

param validate_base_url: bool = True#: If legacy arg openai_api_base is passed in, try to infer if it is a base_url or azure_endpoint and update client params accordingly.

param verbose: bool [Optional]#: Whether to print out response text.

__call__(messages: List[BaseMessage], stop: List[str] | None = None, callbacks: List[BaseCallbackHandler] | BaseCallbackManager | None = None, **kwargs: Any) → BaseMessage#

Deprecated since version langchain-core==0.1.7: Use invoke instead.

Parameters:

messages (List[BaseMessage]) –
stop (List[str] | None) –
callbacks (List[BaseCallbackHandler] | BaseCallbackManager | None) –
kwargs (Any) –

Return type:

BaseMessage

async abatch(inputs: List[Input], config: RunnableConfig | List[RunnableConfig] | None = None, *, return_exceptions: bool = False, **kwargs: Any | None) → List[Output]#

Default implementation runs ainvoke in parallel using asyncio.gather.

The default implementation of batch works well for IO bound runnables.

Subclasses should override this method if they can batch more efficiently; e.g., if the underlying Runnable uses an API which supports a batch mode.

Parameters:

inputs (List[Input]) – A list of inputs to the Runnable.
config (RunnableConfig | List[RunnableConfig] | None) – A config to use when invoking the Runnable. The config supports standard keys like ‘tags’, ‘metadata’ for tracing purposes, ‘max_concurrency’ for controlling how much work to do in parallel, and other keys. Please refer to the RunnableConfig for more details. Defaults to None.
return_exceptions (bool) – Whether to return exceptions instead of raising them. Defaults to False.
kwargs (Any | None) – Additional keyword arguments to pass to the Runnable.

Returns:

A list of outputs from the Runnable.

Return type:

List[Output]

async abatch_as_completed(inputs: Sequence[Input], config: RunnableConfig | Sequence[RunnableConfig] | None = None, *, return_exceptions: bool = False, **kwargs: Any | None) → AsyncIterator[Tuple[int, Output | Exception]]#

Run ainvoke in parallel on a list of inputs, yielding results as they complete.

Parameters:

inputs (Sequence[Input]) – A list of inputs to the Runnable.
config (RunnableConfig | Sequence[RunnableConfig] | None) – A config to use when invoking the Runnable. The config supports standard keys like ‘tags’, ‘metadata’ for tracing purposes, ‘max_concurrency’ for controlling how much work to do in parallel, and other keys. Please refer to the RunnableConfig for more details. Defaults to None. Defaults to None.
return_exceptions (bool) – Whether to return exceptions instead of raising them. Defaults to False.
kwargs (Any | None) – Additional keyword arguments to pass to the Runnable.

Yields:

A tuple of the index of the input and the output from the Runnable.

Return type:

AsyncIterator[Tuple[int, Output | Exception]]

Asynchronously pass a sequence of prompts to a model and return generations.

This method should make use of batched calls for models that expose a batched API.

Use this method when you want to:

take advantage of batched calls,
need more output from the model than just the top generated value,
are building chains that are agnostic to the underlying language model
type (e.g., pure text completion models vs chat models).

Parameters:

messages (List[List[BaseMessage]]) – List of list of messages.
stop (List[str] | None) – Stop words to use when generating. Model output is cut off at the first occurrence of any of these substrings.
callbacks (List[BaseCallbackHandler] | BaseCallbackManager | None) – Callbacks to pass through. Used for executing additional functionality, such as logging or streaming, throughout generation.
**kwargs (Any) – Arbitrary additional keyword arguments. These are usually passed to the model provider API call.
tags (List[str] | None) –
metadata (Dict[str, Any] | None) –
run_name (str | None) –
run_id (UUID | None) –
**kwargs –

Returns:

An LLMResult, which contains a list of candidate Generations for each input: prompt and additional model provider-specific output.

Return type:

LLMResult

async agenerate_prompt(prompts: List[PromptValue], stop: List[str] | None = None, callbacks: List[BaseCallbackHandler] | BaseCallbackManager | None = None, **kwargs: Any) → LLMResult#

Asynchronously pass a sequence of prompts and return model generations.

This method should make use of batched calls for models that expose a batched API.

Use this method when you want to:

take advantage of batched calls,
need more output from the model than just the top generated value,
are building chains that are agnostic to the underlying language model
type (e.g., pure text completion models vs chat models).

Parameters:

prompts (List[PromptValue]) – List of PromptValues. A PromptValue is an object that can be converted to match the format of any language model (string for pure text generation models and BaseMessages for chat models).
stop (List[str] | None) – Stop words to use when generating. Model output is cut off at the first occurrence of any of these substrings.
callbacks (List[BaseCallbackHandler] | BaseCallbackManager | None) – Callbacks to pass through. Used for executing additional functionality, such as logging or streaming, throughout generation.
**kwargs (Any) – Arbitrary additional keyword arguments. These are usually passed to the model provider API call.

Returns:

An LLMResult, which contains a list of candidate Generations for each input: prompt and additional model provider-specific output.

Return type:

LLMResult

async ainvoke(input: LanguageModelInput, config: RunnableConfig | None = None, *, stop: List[str] | None = None, **kwargs: Any) → BaseMessage#

Default implementation of ainvoke, calls invoke from a thread.

The default implementation allows usage of async code even if the Runnable did not implement a native async version of invoke.

Subclasses should override this method if they can run asynchronously.

Parameters:

input (LanguageModelInput) –
config (Optional[RunnableConfig]) –
stop (Optional[List[str]]) –
kwargs (Any) –

Return type:

BaseMessage

async apredict(text: str, *, stop: Sequence[str] | None = None, **kwargs: Any) → str#

Deprecated since version langchain-core==0.1.7: Use ainvoke instead.

Parameters:

text (str) –
stop (Sequence[str] | None) –
kwargs (Any) –

Return type:

str

async apredict_messages(messages: List[BaseMessage], *, stop: Sequence[str] | None = None, **kwargs: Any) → BaseMessage#

Deprecated since version langchain-core==0.1.7: Use ainvoke instead.

Parameters:

messages (List[BaseMessage]) –
stop (Sequence[str] | None) –
kwargs (Any) –

Return type:

BaseMessage

async astream(input: LanguageModelInput, config: RunnableConfig | None = None, *, stop: List[str] | None = None, **kwargs: Any) → AsyncIterator[BaseMessageChunk]#

Default implementation of astream, which calls ainvoke. Subclasses should override this method if they support streaming output.

Parameters:

input (LanguageModelInput) – The input to the Runnable.
config (Optional[RunnableConfig]) – The config to use for the Runnable. Defaults to None.
kwargs (Any) – Additional keyword arguments to pass to the Runnable.
stop (Optional[List[str]]) –

Yields:

The output of the Runnable.

Return type:

AsyncIterator[BaseMessageChunk]

astream_events(input: Any, config: RunnableConfig | None = None, *, version: Literal['v1', 'v2'], include_names: Sequence[str] | None = None, include_types: Sequence[str] | None = None, include_tags: Sequence[str] | None = None, exclude_names: Sequence[str] | None = None, exclude_types: Sequence[str] | None = None, exclude_tags: Sequence[str] | None = None, **kwargs: Any) → AsyncIterator[StandardStreamEvent | CustomStreamEvent]#

Beta

This API is in beta and may change in the future.

Generate a stream of events.

Use to create an iterator over StreamEvents that provide real-time information about the progress of the Runnable, including StreamEvents from intermediate results.

A StreamEvent is a dictionary with the following schema:

event: str - Event names are of the
format: on_[runnable_type]_(start|stream|end).
name: str - The name of the Runnable that generated the event.
run_id: str - randomly generated ID associated with the given execution of
the Runnable that emitted the event. A child Runnable that gets invoked as part of the execution of a parent Runnable is assigned its own unique ID.
parent_ids: List[str] - The IDs of the parent runnables that
generated the event. The root Runnable will have an empty list. The order of the parent IDs is from the root to the immediate parent. Only available for v2 version of the API. The v1 version of the API will return an empty list.
tags: Optional[List[str]] - The tags of the Runnable that generated
the event.
metadata: Optional[Dict[str, Any]] - The metadata of the Runnable
that generated the event.
data: Dict[str, Any]

Below is a table that illustrates some evens that might be emitted by various chains. Metadata fields have been omitted from the table for brevity. Chain definitions have been included after the table.

ATTENTION This reference table is for the V2 version of the schema.

event	name	chunk	input	output
on_chat_model_start	[model name]		{“messages”: [[SystemMessage, HumanMessage]]}
on_chat_model_stream	[model name]	AIMessageChunk(content=”hello”)
on_chat_model_end	[model name]		{“messages”: [[SystemMessage, HumanMessage]]}	AIMessageChunk(content=”hello world”)
on_llm_start	[model name]		{‘input’: ‘hello’}
on_llm_stream	[model name]	‘Hello’
on_llm_end	[model name]		‘Hello human!’
on_chain_start	format_docs
on_chain_stream	format_docs	“hello world!, goodbye world!”
on_chain_end	format_docs		[Document(…)]	“hello world!, goodbye world!”
on_tool_start	some_tool		{“x”: 1, “y”: “2”}
on_tool_end	some_tool			{“x”: 1, “y”: “2”}
on_retriever_start	[retriever name]		{“query”: “hello”}
on_retriever_end	[retriever name]		{“query”: “hello”}	[Document(…), ..]
on_prompt_start	[template_name]		{“question”: “hello”}
on_prompt_end	[template_name]		{“question”: “hello”}	ChatPromptValue(messages: [SystemMessage, …])

In addition to the standard events, users can also dispatch custom events (see example below).

Custom events will be only be surfaced with in the v2 version of the API!

A custom event has following format:

Attribute	Type	Description
name	str	A user defined name for the event.
data	Any	The data associated with the event. This can be anything, though we suggest making it JSON serializable.

Here are declarations associated with the standard events shown above:

format_docs:

def format_docs(docs: List[Document]) -> str:
    '''Format the docs.'''
    return ", ".join([doc.page_content for doc in docs])

format_docs = RunnableLambda(format_docs)

some_tool:

@tool
def some_tool(x: int, y: str) -> dict:
    '''Some_tool.'''
    return {"x": x, "y": y}

prompt:

template = ChatPromptTemplate.from_messages(
    [("system", "You are Cat Agent 007"), ("human", "{question}")]
).with_config({"run_name": "my_template", "tags": ["my_template"]})

Example:

from langchain_core.runnables import RunnableLambda

async def reverse(s: str) -> str:
    return s[::-1]

chain = RunnableLambda(func=reverse)

events = [
    event async for event in chain.astream_events("hello", version="v2")
]

# will produce the following events (run_id, and parent_ids
# has been omitted for brevity):
[
    {
        "data": {"input": "hello"},
        "event": "on_chain_start",
        "metadata": {},
        "name": "reverse",
        "tags": [],
    },
    {
        "data": {"chunk": "olleh"},
        "event": "on_chain_stream",
        "metadata": {},
        "name": "reverse",
        "tags": [],
    },
    {
        "data": {"output": "olleh"},
        "event": "on_chain_end",
        "metadata": {},
        "name": "reverse",
        "tags": [],
    },
]

Example: Dispatch Custom Event

from langchain_core.callbacks.manager import (
    adispatch_custom_event,
)
from langchain_core.runnables import RunnableLambda, RunnableConfig
import asyncio


async def slow_thing(some_input: str, config: RunnableConfig) -> str:
    """Do something that takes a long time."""
    await asyncio.sleep(1) # Placeholder for some slow operation
    await adispatch_custom_event(
        "progress_event",
        {"message": "Finished step 1 of 3"},
        config=config # Must be included for python < 3.10
    )
    await asyncio.sleep(1) # Placeholder for some slow operation
    await adispatch_custom_event(
        "progress_event",
        {"message": "Finished step 2 of 3"},
        config=config # Must be included for python < 3.10
    )
    await asyncio.sleep(1) # Placeholder for some slow operation
    return "Done"

slow_thing = RunnableLambda(slow_thing)

async for event in slow_thing.astream_events("some_input", version="v2"):
    print(event)

Parameters:

input (Any) – The input to the Runnable.
config (RunnableConfig | None) – The config to use for the Runnable.
version (Literal['v1', 'v2']) – The version of the schema to use either v2 or v1. Users should use v2. v1 is for backwards compatibility and will be deprecated in 0.4.0. No default will be assigned until the API is stabilized. custom events will only be surfaced in v2.
include_names (Sequence[str] | None) – Only include events from runnables with matching names.
include_types (Sequence[str] | None) – Only include events from runnables with matching types.
include_tags (Sequence[str] | None) – Only include events from runnables with matching tags.
exclude_names (Sequence[str] | None) – Exclude events from runnables with matching names.
exclude_types (Sequence[str] | None) – Exclude events from runnables with matching types.
exclude_tags (Sequence[str] | None) – Exclude events from runnables with matching tags.
kwargs (Any) – Additional keyword arguments to pass to the Runnable. These will be passed to astream_log as this implementation of astream_events is built on top of astream_log.

Yields:

An async stream of StreamEvents.

Raises:

NotImplementedError – If the version is not v1 or v2.

Return type:

AsyncIterator[StandardStreamEvent | CustomStreamEvent]

batch(inputs: List[Input], config: RunnableConfig | List[RunnableConfig] | None = None, *, return_exceptions: bool = False, **kwargs: Any | None) → List[Output]#

Default implementation runs invoke in parallel using a thread pool executor.

The default implementation of batch works well for IO bound runnables.

Subclasses should override this method if they can batch more efficiently; e.g., if the underlying Runnable uses an API which supports a batch mode.

Parameters:

inputs (List[Input]) –
config (RunnableConfig | List[RunnableConfig] | None) –
return_exceptions (bool) –
kwargs (Any | None) –

Return type:

List[Output]

batch_as_completed(inputs: Sequence[Input], config: RunnableConfig | Sequence[RunnableConfig] | None = None, *, return_exceptions: bool = False, **kwargs: Any | None) → Iterator[Tuple[int, Output | Exception]]#

Run invoke in parallel on a list of inputs, yielding results as they complete.

Parameters:

inputs (Sequence[Input]) –
config (RunnableConfig | Sequence[RunnableConfig] | None) –
return_exceptions (bool) –
kwargs (Any | None) –

Return type:

Iterator[Tuple[int, Output | Exception]]

Bind functions (and other objects) to this chat model.

Assumes model is compatible with OpenAI function-calling API.

NOTE: Using bind_tools is recommended instead, as the functions and: function_call request parameters are officially marked as deprecated by OpenAI.

Parameters:

functions (Sequence[Dict[str, Any] | Type[BaseModel] | Callable | BaseTool]) – A list of function definitions to bind to this chat model. Can be a dictionary, pydantic model, or callable. Pydantic models and callables will be automatically converted to their schema dictionary representation.
function_call (_FunctionCall | str | Literal['auto', 'none'] | None) – Which function to require the model to call. Must be the name of the single provided function or “auto” to automatically determine which function to call (if any).
**kwargs (Any) – Any additional parameters to pass to the Runnable constructor.

Return type:

Bind tool-like objects to this chat model.

Assumes model is compatible with OpenAI tool-calling API.

Changed in version 0.1.21: Support for strict argument added.

Parameters:

tools (Sequence[Dict[str, Any] | Type | Callable | BaseTool]) – A list of tool definitions to bind to this chat model. Supports any tool definition handled by langchain_core.utils.function_calling.convert_to_openai_tool().
tool_choice (dict | str | Literal['auto', 'none', 'any', 'required'] | bool | None) –
Which tool to require the model to call. Options are:
- str of the form "<<tool_name>>": calls <<tool_name>> tool.
- "auto": automatically selects a tool (including no tool).
- "none": does not call a tool.
- "any" or "required" or True: force at least one tool to be called.
- dict of the form {"type": "function", "function": {"name": <<tool_name>>}}: calls <<tool_name>> tool.
- False or None: no effect, default OpenAI behavior.
strict –
If True, model output is guaranteed to exactly match the JSON Schema provided in the tool definition. If True, the input schema will be validated according to https://platform.openai.com/docs/guides/structured-outputs/supported-schemas. If False, input schema will not be validated and model output will not be validated. If None, strict argument will not be passed to the model.

New in version 0.1.21.
kwargs (Any) – Any additional parameters are passed directly to self.bind(**kwargs).

Return type:

call_as_llm(message: str, stop: List[str] | None = None, **kwargs: Any) → str#

Deprecated since version langchain-core==0.1.7: Use invoke instead.

Parameters:

message (str) –
stop (List[str] | None) –
kwargs (Any) –

Return type:

str

configurable_alternatives(which: ConfigurableField, *, default_key: str = 'default', prefix_keys: bool = False, **kwargs: Runnable[Input, Output] | Callable[[], Runnable[Input, Output]]) → RunnableSerializable[Input, Output]#

Configure alternatives for Runnables that can be set at runtime.

Parameters:

which (ConfigurableField) – The ConfigurableField instance that will be used to select the alternative.
default_key (str) – The default key to use if no alternative is selected. Defaults to “default”.
prefix_keys (bool) – Whether to prefix the keys with the ConfigurableField id. Defaults to False.
**kwargs (Runnable[Input, Output] | Callable[[], Runnable[Input, Output]]) – A dictionary of keys to Runnable instances or callables that return Runnable instances.

Returns:

A new Runnable with the alternatives configured.

Return type:

RunnableSerializable[Input, Output]

from langchain_anthropic import ChatAnthropic
from langchain_core.runnables.utils import ConfigurableField
from langchain_openai import ChatOpenAI

model = ChatAnthropic(
    model_name="claude-3-sonnet-20240229"
).configurable_alternatives(
    ConfigurableField(id="llm"),
    default_key="anthropic",
    openai=ChatOpenAI()
)

# uses the default model ChatAnthropic
print(model.invoke("which organization created you?").content)

# uses ChatOpenAI
print(
    model.with_config(
        configurable={"llm": "openai"}
    ).invoke("which organization created you?").content
)

configurable_fields(**kwargs: ConfigurableField | ConfigurableFieldSingleOption | ConfigurableFieldMultiOption) → RunnableSerializable[Input, Output]#

Configure particular Runnable fields at runtime.

Parameters:: **kwargs (ConfigurableField | ConfigurableFieldSingleOption | ConfigurableFieldMultiOption) – A dictionary of ConfigurableField instances to configure.
Returns:: A new Runnable with the fields configured.
Return type:: RunnableSerializable[Input, Output]

from langchain_core.runnables import ConfigurableField
from langchain_openai import ChatOpenAI

model = ChatOpenAI(max_tokens=20).configurable_fields(
    max_tokens=ConfigurableField(
        id="output_token_number",
        name="Max tokens in the output",
        description="The maximum number of tokens in the output",
    )
)

# max_tokens = 20
print(
    "max_tokens_20: ",
    model.invoke("tell me something about chess").content
)

# max_tokens = 200
print("max_tokens_200: ", model.with_config(
    configurable={"output_token_number": 200}
    ).invoke("tell me something about chess").content
)

Pass a sequence of prompts to the model and return model generations.

This method should make use of batched calls for models that expose a batched API.

Use this method when you want to:

take advantage of batched calls,
need more output from the model than just the top generated value,
are building chains that are agnostic to the underlying language model
type (e.g., pure text completion models vs chat models).

Parameters:

messages (List[List[BaseMessage]]) – List of list of messages.
stop (List[str] | None) – Stop words to use when generating. Model output is cut off at the first occurrence of any of these substrings.
callbacks (List[BaseCallbackHandler] | BaseCallbackManager | None) – Callbacks to pass through. Used for executing additional functionality, such as logging or streaming, throughout generation.
**kwargs (Any) – Arbitrary additional keyword arguments. These are usually passed to the model provider API call.
tags (List[str] | None) –
metadata (Dict[str, Any] | None) –
run_name (str | None) –
run_id (UUID | None) –
**kwargs –

Returns:

An LLMResult, which contains a list of candidate Generations for each input: prompt and additional model provider-specific output.

Return type:

LLMResult

generate_prompt(prompts: List[PromptValue], stop: List[str] | None = None, callbacks: List[BaseCallbackHandler] | BaseCallbackManager | None = None, **kwargs: Any) → LLMResult#

Pass a sequence of prompts to the model and return model generations.

This method should make use of batched calls for models that expose a batched API.

Use this method when you want to:

take advantage of batched calls,
need more output from the model than just the top generated value,
are building chains that are agnostic to the underlying language model
type (e.g., pure text completion models vs chat models).

Parameters:

prompts (List[PromptValue]) – List of PromptValues. A PromptValue is an object that can be converted to match the format of any language model (string for pure text generation models and BaseMessages for chat models).
stop (List[str] | None) – Stop words to use when generating. Model output is cut off at the first occurrence of any of these substrings.
callbacks (List[BaseCallbackHandler] | BaseCallbackManager | None) – Callbacks to pass through. Used for executing additional functionality, such as logging or streaming, throughout generation.
**kwargs (Any) – Arbitrary additional keyword arguments. These are usually passed to the model provider API call.

Returns:

An LLMResult, which contains a list of candidate Generations for each input: prompt and additional model provider-specific output.

Return type:

LLMResult

get_num_tokens(text: str) → int#

Get the number of tokens present in the text.

Useful for checking if an input fits in a model’s context window.

Parameters:: text (str) – The string input to tokenize.
Returns:: The integer number of tokens in the text.
Return type:: int

get_num_tokens_from_messages(messages: List[BaseMessage]) → int#

Calculate num tokens for gpt-3.5-turbo and gpt-4 with tiktoken package.

Requirements: You must have the pillow installed if you want to count image tokens if you are specifying the image as a base64 string, and you must have both pillow and httpx installed if you are specifying the image as a URL. If these aren’t installed image inputs will be ignored in token counting.

OpenAI reference: openai/openai-cookbook main/examples/How_to_format_inputs_to_ChatGPT_models.ipynb

Parameters:: messages (List[BaseMessage]) –
Return type:: int

get_token_ids(text: str) → List[int]#

Get the tokens present in the text with tiktoken package.

Parameters:: text (str) –
Return type:: List[int]

invoke(input: LanguageModelInput, config: RunnableConfig | None = None, *, stop: List[str] | None = None, **kwargs: Any) → BaseMessage#

Transform a single input into an output. Override to implement.

Parameters:

input (LanguageModelInput) – The input to the Runnable.
config (Optional[RunnableConfig]) – A config to use when invoking the Runnable. The config supports standard keys like ‘tags’, ‘metadata’ for tracing purposes, ‘max_concurrency’ for controlling how much work to do in parallel, and other keys. Please refer to the RunnableConfig for more details.
stop (Optional[List[str]]) –
kwargs (Any) –

Returns:

The output of the Runnable.

Return type:

BaseMessage

predict(text: str, *, stop: Sequence[str] | None = None, **kwargs: Any) → str#

Deprecated since version langchain-core==0.1.7: Use invoke instead.

Parameters:

text (str) –
stop (Sequence[str] | None) –
kwargs (Any) –

Return type:

str

predict_messages(messages: List[BaseMessage], *, stop: Sequence[str] | None = None, **kwargs: Any) → BaseMessage#

Deprecated since version langchain-core==0.1.7: Use invoke instead.

Parameters:

messages (List[BaseMessage]) –
stop (Sequence[str] | None) –
kwargs (Any) –

Return type:

BaseMessage

stream(input: LanguageModelInput, config: RunnableConfig | None = None, *, stop: List[str] | None = None, **kwargs: Any) → Iterator[BaseMessageChunk]#

Default implementation of stream, which calls invoke. Subclasses should override this method if they support streaming output.

Parameters:

input (LanguageModelInput) – The input to the Runnable.
config (Optional[RunnableConfig]) – The config to use for the Runnable. Defaults to None.
kwargs (Any) – Additional keyword arguments to pass to the Runnable.
stop (Optional[List[str]]) –

Yields:

The output of the Runnable.

Return type:

Iterator[BaseMessageChunk]

to_json() → SerializedConstructor | SerializedNotImplemented#

Serialize the Runnable to JSON.

Returns:: A JSON-serializable representation of the Runnable.
Return type:: SerializedConstructor | SerializedNotImplemented

Model wrapper that returns outputs formatted to match the given schema.

Args:

schema:

The output schema. Can be passed in as:

an OpenAI function/tool schema,
a JSON Schema,
a TypedDict class,
or a Pydantic class.

If schema is a Pydantic class then the model output will be a Pydantic instance of that class, and the model-generated fields will be validated by the Pydantic class. Otherwise the model output will be a dict and will not be validated. See langchain_core.utils.function_calling.convert_to_openai_tool() for more on how to properly specify types and descriptions of schema fields when specifying a Pydantic or TypedDict class.

method:

The method for steering model generation, either “function_calling” or “json_mode”. If “function_calling” then the schema will be converted to an OpenAI function and the returned model will make use of the function-calling API. If “json_mode” then OpenAI’s JSON mode will be used. Note that if using “json_mode” then you must include instructions for formatting the output into the desired schema into the model call.

include_raw:

If False then only the parsed structured output is returned. If an error occurs during model output parsing it will be raised. If True then both the raw model response (a BaseMessage) and the parsed model response will be returned. If an error occurs during output parsing it will be caught and returned as well. The final output is always a dict with keys “raw”, “parsed”, and “parsing_error”.

Returns:

A Runnable that takes same inputs as a langchain_core.language_models.chat.BaseChatModel.

If include_raw is False and schema is a Pydantic class, Runnable outputs an instance of schema (i.e., a Pydantic object).

Otherwise, if include_raw is False then Runnable outputs a dict.

If include_raw is True, then Runnable outputs a dict with keys:

"raw": BaseMessage
"parsed": None if there was a parsing error, otherwise the type depends on the schema as described above.
"parsing_error": Optional[BaseException]

Example: schema=Pydantic class, method=”function_calling”, include_raw=False:

from typing import Optional

from langchain_openai import AzureChatOpenAI
from langchain_core.pydantic_v1 import BaseModel, Field


class AnswerWithJustification(BaseModel):
    '''An answer to the user question along with justification for the answer.'''

    answer: str
    # If we provide default values and/or descriptions for fields, these will be passed
    # to the model. This is an important part of improving a model's ability to
    # correctly return structured outputs.
    justification: Optional[str] = Field(
        default=None, description="A justification for the answer."
    )


llm = AzureChatOpenAI(model="gpt-3.5-turbo-0125", temperature=0)
structured_llm = llm.with_structured_output(AnswerWithJustification)

structured_llm.invoke(
    "What weighs more a pound of bricks or a pound of feathers"
)

# -> AnswerWithJustification(
#     answer='They weigh the same',
#     justification='Both a pound of bricks and a pound of feathers weigh one pound. The weight is the same, but the volume or density of the objects may differ.'
# )

Example: schema=Pydantic class, method=”function_calling”, include_raw=True:

from langchain_openai import AzureChatOpenAI
from langchain_core.pydantic_v1 import BaseModel


class AnswerWithJustification(BaseModel):
    '''An answer to the user question along with justification for the answer.'''

    answer: str
    justification: str


llm = AzureChatOpenAI(model="gpt-3.5-turbo-0125", temperature=0)
structured_llm = llm.with_structured_output(
    AnswerWithJustification, include_raw=True
)

structured_llm.invoke(
    "What weighs more a pound of bricks or a pound of feathers"
)
# -> {
#     'raw': AIMessage(content='', additional_kwargs={'tool_calls': [{'id': 'call_Ao02pnFYXD6GN1yzc0uXPsvF', 'function': {'arguments': '{"answer":"They weigh the same.","justification":"Both a pound of bricks and a pound of feathers weigh one pound. The weight is the same, but the volume or density of the objects may differ."}', 'name': 'AnswerWithJustification'}, 'type': 'function'}]}),
#     'parsed': AnswerWithJustification(answer='They weigh the same.', justification='Both a pound of bricks and a pound of feathers weigh one pound. The weight is the same, but the volume or density of the objects may differ.'),
#     'parsing_error': None
# }

Example: schema=TypedDict class, method=”function_calling”, include_raw=False:

# IMPORTANT: If you are using Python <=3.8, you need to import Annotated
# from typing_extensions, not from typing.
from typing_extensions import Annotated, TypedDict

from langchain_openai import AzureChatOpenAI


class AnswerWithJustification(TypedDict):
    '''An answer to the user question along with justification for the answer.'''

    answer: str
    justification: Annotated[
        Optional[str], None, "A justification for the answer."
    ]


llm = AzureChatOpenAI(model="gpt-3.5-turbo-0125", temperature=0)
structured_llm = llm.with_structured_output(AnswerWithJustification)

structured_llm.invoke(
    "What weighs more a pound of bricks or a pound of feathers"
)
# -> {
#     'answer': 'They weigh the same',
#     'justification': 'Both a pound of bricks and a pound of feathers weigh one pound. The weight is the same, but the volume and density of the two substances differ.'
# }

Example: schema=OpenAI function schema, method=”function_calling”, include_raw=False:

 from langchain_openai import AzureChatOpenAI

 oai_schema = {
     'name': 'AnswerWithJustification',
     'description': 'An answer to the user question along with justification for the answer.',
     'parameters': {
         'type': 'object',
         'properties': {
             'answer': {'type': 'string'},
             'justification': {'description': 'A justification for the answer.', 'type': 'string'}
         },
        'required': ['answer']
    }
}

 llm = AzureChatOpenAI(model="gpt-3.5-turbo-0125", temperature=0)
 structured_llm = llm.with_structured_output(oai_schema)

 structured_llm.invoke(
     "What weighs more a pound of bricks or a pound of feathers"
 )
 # -> {
 #     'answer': 'They weigh the same',
 #     'justification': 'Both a pound of bricks and a pound of feathers weigh one pound. The weight is the same, but the volume and density of the two substances differ.'
 # }

Example: schema=Pydantic class, method=”json_mode”, include_raw=True:

from langchain_openai import AzureChatOpenAI
from langchain_core.pydantic_v1 import BaseModel

class AnswerWithJustification(BaseModel):
    answer: str
    justification: str

llm = AzureChatOpenAI(model="gpt-3.5-turbo-0125", temperature=0)
structured_llm = llm.with_structured_output(
    AnswerWithJustification,
    method="json_mode",
    include_raw=True
)

structured_llm.invoke(
    "Answer the following question. "
    "Make sure to return a JSON blob with keys 'answer' and 'justification'.

“

“What’s heavier a pound of bricks or a pound of feathers?”

) # -> { # ‘raw’: AIMessage(content=’{

“answer”: “They are both the same weight.”, “justification”: “Both a pound of bricks and a pound of feathers weigh one pound. The difference lies in the volume and density of the materials, not the weight.”

}’),

# ‘parsed’: AnswerWithJustification(answer=’They are both the same weight.’, justification=’Both a pound of bricks and a pound of feathers weigh one pound. The difference lies in the volume and density of the materials, not the weight.’), # ‘parsing_error’: None # }

Example: schema=None, method=”json_mode”, include_raw=True:

structured_llm = llm.with_structured_output(method="json_mode", include_raw=True)

structured_llm.invoke(
    "Answer the following question. "
    "Make sure to return a JSON blob with keys 'answer' and 'justification'.

“: “What’s heavier a pound of bricks or a pound of feathers?”

) # -> { # ‘raw’: AIMessage(content=’{

“answer”: “They are both the same weight.”, “justification”: “Both a pound of bricks and a pound of feathers weigh one pound. The difference lies in the volume and density of the materials, not the weight.”
}’),: # ‘parsed’: { # ‘answer’: ‘They are both the same weight.’, # ‘justification’: ‘Both a pound of bricks and a pound of feathers weigh one pound. The difference lies in the volume and density of the materials, not the weight.’ # }, # ‘parsing_error’: None # }

Examples using AzureChatOpenAI