Fireworks AI API

fireworks.ai

Provides API access to run inference on open-source LLMs, vision, image, and audio models, plus fine-tuning and dedicated deployments.

The Fireworks AI API API has usage based pricing; offers a free tier ($1 in free credits to get started); rate limits: High rate limits on serverless; higher rate limits available with on-demand and enterprise (specific numbers not documented). Verified 2026-07-13 with per-field source links.

AI & Machine Learning Verified 2026-07-13 medium confidence

Base URL

not documented null

no source

Auth

not documented null

no source

Free tier

$1 in free credits to get started

fireworks.ai ↗

Pricing model

usage_based

fireworks.ai ↗

Pricing

Serverless pay-per-token (e.g. GLM 5.2 $1.4/M input, $4.4/M output); embeddings from $0.008/1M input tokens; on-demand GPU $7/hr (H100/H200), $10/hr (B200), $12/hr (B300); fine tuning from $0.50/1M training tokens; cached input and batch inference at 50%

fireworks.ai ↗

Rate limits

High rate limits on serverless; higher rate limits available with on-demand and enterprise (specific numbers not documented)

fireworks.ai ↗

OpenAPI spec

not documented null

no source

MCP server

not documented null

no source