ChatFireworks

Internal fireworks.Fireworks().chat.completions resource.

Constructed with max_retries=0 so retries are owned by _create_retry_decorator (which surfaces each attempt to the LangChain run_manager). Callers reaching for this directly should set their own retry layer.

Internal fireworks.AsyncFireworks().chat.completions resource.

Constructed with max_retries=0; see client.

Model name to use.

Same as model_name.

What sampling temperature to use.

Default stop sequences.

Holds any model parameters valid for create call not explicitly specified.

Fireworks API key.

Automatically read from env variable FIREWORKS_API_KEY if not provided.

Base URL path for API requests, leave blank if not using a proxy or service emulator.

Timeout for requests to Fireworks completion API. Can be float, httpx.Timeout or None.

Whether to stream the results or not.

Whether to include usage metadata in streaming output.

If True, a final empty-content chunk carrying usage_metadata is emitted during the stream. Set to False if the upstream model/proxy rejects stream_options, or pass stream_options explicitly via model_kwargs or a runtime kwarg to override.

Number of chat completions to generate for each prompt.

Maximum number of tokens to generate.

Maximum number of retries after the initial attempt when generating.

Retries use exponential backoff and trigger on transient errors: RateLimitError, APIConnectionError (including its APITimeoutError subclass), 5xx responses (including those that surface as httpx.HTTPStatusError rather than typed SDK errors), and underlying transport errors (httpx.TimeoutException, httpx.TransportError). A value of None or 0 disables retries.

Get the namespace of the LangChain object.

Return whether this model can be serialized by LangChain.

Build extra kwargs from additional params that were passed in.

Validate that api key and python package exists in environment.

Close the underlying sync HTTP client.

After calling, sync invocations on this model will raise. Async invocations remain available until aclose() is also called. Safe to call multiple times.

Close the underlying async HTTP client.

Releases the aiohttp-backed connector that the 1.x SDK uses by default. Without this, transient ChatFireworks instances can leak an Unclosed connector warning at GC if the event loop has already stopped. Safe to call multiple times.

Bind tool-like objects to this chat model.

Assumes model is compatible with Fireworks tool-calling API.

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