Quickstart

Connect a coding harness, application, or LLM gateway to Fireworks and make your first Nexus request.

Connect through a coding harness, an API or SDK, or an LLM gateway. Harnesses and direct API clients use the same Fireworks model IDs and account controls. Gateway setup varies by provider.

Coding harness

Install FireConnect, sign in, and connect the harness you already use:

curl -fsSL https://fireconnect.fireworks.ai/install.sh | bash

fireconnect login
fireconnect claude

Or use the raw GitHub URL instead: curl -fsSL https://raw.githubusercontent.com/fw-ai/fireconnect/main/install.sh | bash.

Restart Claude Code, then open /model. FireConnect automatically adds coding-ready open models and auto. Replace claude with another harness from Coding Harnesses.

API or SDK

Point an OpenAI-compatible client at Fireworks and choose a model:

pip install openai
export FIREWORKS_API_KEY="YOUR_FIREWORKS_API_KEY"

import os

from openai import OpenAI

client = OpenAI(
    api_key=os.environ["FIREWORKS_API_KEY"],
    base_url="https://api.fireworks.ai/inference/v1",
)

response = client.responses.create(
    model="glm-latest",
    input="Review this pull request.",
)

print(response.output_text)

Fireworks also supports Chat Completions and Anthropic Messages. See APIs and SDKs.

LLM gateway

Keep your existing gateway and configure its Fireworks provider. For LiteLLM, create config.yaml:

model_list:
  - model_name: glm-latest
    litellm_params:
      model: fireworks_ai/glm-latest
      api_key: os.environ/FIREWORKS_API_KEY
      api_base: https://api.fireworks.ai/inference/v1

Start the proxy:

pip install "litellm[proxy]"
export FIREWORKS_API_KEY="YOUR_FIREWORKS_API_KEY"
litellm --config config.yaml

In another terminal, send a request:

curl http://localhost:4000/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "glm-latest",
    "messages": [{"role": "user", "content": "Say pong in one word."}]
  }'

Provider names, key configuration, and model catalogs vary by gateway. See LLM Gateways.

Choose models or manage usage and spend

Open Models

Use auto, follow a model family, choose a fast tier, or pin a version.

FireRouter

Let one model ID choose a model for each new user turn.

Usage and Cost

See session estimates and Fireworks model usage by model, user, or API key.

Spend Limits

Set per-user caps for supported Fireworks serverless models, with account defaults, group limits, and user exceptions.