Fireworks LangChain Integration

Warnings

• breaking LangChain, as a rapidly evolving ecosystem, frequently undergoes API changes, especially with major version bumps of `langchain-core`. Code from older tutorials or examples may use deprecated patterns and require migration to newer interfaces (e.g., LCEL).
  Fix: Refer to the official LangChain documentation and migration guides for the latest API usage. Ensure `langchain-core` is updated to a compatible version.
• gotcha The `FIREWORKS_API_KEY` environment variable is the primary method for authenticating with Fireworks AI. Failure to set this variable correctly or pass it directly to the model constructor will result in authentication errors.
  Fix: Set the `FIREWORKS_API_KEY` environment variable or pass `fireworks_api_key='your-key'` directly during model instantiation.
• gotcha The library explicitly requires Python versions `>=3.10.0` and `<4.0.0`. Using unsupported Python versions (e.g., older than 3.10 or Python 4.x) may lead to installation failures or runtime incompatibilities.
  Fix: Ensure your Python environment is running a compatible version (3.10, 3.11, or 3.12).
• gotcha When using advanced features like tool calling or structured output, some specific Fireworks models have historically exhibited issues with retaining `reasoning_content` in multi-turn conversations, potentially leading to incorrect responses or hallucinations.
  Fix: Thoroughly test tool use and structured output with your chosen model. Monitor Fireworks AI's release notes and LangChain's changelog for model-specific fixes and improvements. Ensure the latest version of `langchain-fireworks` is installed, as improvements for structured output have been released.

Install

• pip install langchain-fireworks Install latest version

Imports

• ChatFireworks

  from langchain_fireworks import ChatFireworks

• FireworksEmbeddings

  from langchain_fireworks import FireworksEmbeddings

Quickstart

Demonstrates how to initialize `ChatFireworks` with a specific model and make a simple chat completion call, ensuring the API key is provided, preferably via an environment variable. The model `llama-v3p3-70b-instruct` is an example; refer to Fireworks AI for the latest model IDs.

import os
from langchain_fireworks import ChatFireworks
from langchain_core.messages import HumanMessage, SystemMessage

# Ensure FIREWORKS_API_KEY environment variable is set.
# For a real application, load your API key securely.
# os.environ["FIREWORKS_API_KEY"] = "YOUR_FIREWORKS_API_KEY"

llm = ChatFireworks(
    model="accounts/fireworks/models/llama-v3p3-70b-instruct",
    temperature=0,
    max_tokens=None,
    fireworks_api_key=os.environ.get("FIREWORKS_API_KEY", "") # Fallback for quickstart if not set
)

messages = [
    SystemMessage(content="You are a helpful assistant."),
    HumanMessage(content="What is the capital of France?"),
]

response = llm.invoke(messages)
print(response.content)