Introducing Parallel Decisions API: the first decision engine with world knowledge built in. Learn more.[Learn more.]
Fireworks logo

## What you can do

  • - Let Fireworks-hosted open models search the web with Parallel when a question needs current information.
  • - Return answers that cite the source URLs from Parallel Search results.

## How it works

Fireworks serves open models through an OpenAI-compatible chat completions API. Give the model a web_search tool backed by Parallel Search, and it returns a tool call with a search objective and a few keyword queries when it needs fresh information.

Your code sends the objective and queries to Parallel Search and returns the results to the model as a tool message. The model can search more than once, then writes a final answer from the results and lists the URLs it used.

## Set up in three steps

  1. 1. Export a Fireworks API key as FIREWORKS_API_KEY and a Parallel API key as PARALLEL_API_KEY.
  2. 2. Point the OpenAI SDK at the Fireworks base URL and define a web_search tool that sends its objective and search queries to Parallel Search.
  3. 3. Ask a question with a Fireworks model that supports function calling, return each search result to the model, and verify the answer lists its source URLs.
Add Parallel to Fireworks[Add Parallel to Fireworks]