DurableChatModel

Package: flyteplugins.agents.deepagents

Wrap a BaseChatModel so each model turn is durable and replayable.

_agenerate (async) delegates to the inner model and records the turn via durable_step. Pass an instance as the deep agent’s model — create_deep_agent(model=DurableChatModel(inner=model), ...) — or as a subagent’s model; bind_tools and other capabilities are delegated to the inner model so tool-calling behaves exactly as the inner model does.

Durability is best-effort: if anything in the durable path raises, the turn falls back to a direct inner call so a run is never broken by it.

Parameters

class DurableChatModel(
    name: str | None = None,
    cache: langchain_core.caches.BaseCache | bool | None = None,
    verbose: bool = _get_verbosity(),
    callbacks: list[langchain_core.callbacks.base.BaseCallbackHandler] | langchain_core.callbacks.base.BaseCallbackManager | None = None,
    tags: list[str] | None = None,
    metadata: dict[str, typing.Any] | None = None,
    custom_get_token_ids: collections.abc.Callable[[str], list[int]] | None = None,
    rate_limiter: langchain_core.rate_limiters.BaseRateLimiter | None = None,
    disable_streaming: typing.Union[bool, typing.Literal['tool_calling']] = False,
    output_version: str | None = get_from_env_fn(),
    profile: langchain_core.language_models.model_profile.ModelProfile | None = None,
    inner: langchain_core.language_models.chat_models.BaseChatModel,
)

Methods

bind_tools()

def bind_tools(
    tools: typing.Sequence[typing.Any],
    **kwargs: typing.Any,
) -> 'Runnable'

Format tools via the inner model, but bind them to this wrapper.

The inner model knows how to convert tools into its provider format; we reuse that, then re-bind the resulting kwargs to self so the runnable the deep agent invokes still routes generation through the durable override (rather than the inner model directly).

get_num_tokens()

def 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.

This should be overridden by model-specific implementations to provide accurate token counts via model-specific tokenizers.

Returns: The integer number of tokens in the text.

get_num_tokens_from_messages()

def get_num_tokens_from_messages(
    messages,
    tools = None,
) -> int

Get the number of tokens in the messages.

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

This should be overridden by model-specific implementations to provide accurate token counts via model-specific tokenizers.

Returns: The sum of the number of tokens across the messages.