Skip to content
SocialAtoZ

Fireworks AI

Verified

Fast, production-ready inference and fine-tuning for open source AI models

AI Development Platform
–

Not yet rated. Be the first to review Fireworks AI.

Visit website Request pricing Write a review
Fireworks AI screenshot See all screenshots
  • Deployment Cloud Based
  • Starting price Pay as you go
  • Free trial Available
  • Best for Startups, SMEs, Enterprises

What is Fireworks AI?

Fireworks AI provides serverless API access to more than 400 open-source models across text, vision, image, and audio, along with fine-tuning and dedicated GPU deployment options. Built on proprietary optimization technology including the FireAttention CUDA kernel and speculative decoding, the platform advertises sub-100ms first-token latency and is used in production by companies such as Uber, DoorDash, and Notion.

Founded in 2022 by former Meta PyTorch engineers and based in Redwood City, California, Fireworks AI charges per token for serverless inference with new accounts receiving $1 in free credits, per-GPU-hour rates for dedicated deployments on H100, H200, B200, and B300 hardware, and custom Enterprise pricing with SOC 2, GDPR, and HIPAA compliance.

Key Features of Fireworks AI

Fireworks AI lists 8 documented features, including Serverless API access to 400+ open source models, Sub-100ms first-token latency and LoRA, full fine-tuning, and reinforcement fine-tuning. The list below covers what the product does rather than how it is marketed.

  • Serverless API access to 400+ open source models
  • Sub-100ms first-token latency
  • LoRA, full fine-tuning, and reinforcement fine-tuning
  • Dedicated GPU deployments (H100, H200, B200, B300)
  • 50% discount on cached input tokens
  • Batch inference at 50% of serverless pricing
  • FireAttention custom CUDA kernel and speculative decoding
  • SOC 2 Type II, GDPR, and HIPAA compliance

Fireworks AI Pricing

Fireworks AI lists 3 plans without published figures. Pricing is quoted on request.

Developer

Pay as you go

$1 free starter credit

Request pricing

Per-token serverless inference pricing; varies by model, e.g. gpt-oss-120b at $0.20/1M tokens

On-Demand Deployments

From $7.00/GPU-hour

Dedicated GPU

Request pricing

H100 and H200 at $7/hr, B200 at $10/hr, B300 at $12/hr

Enterprise

Custom

Request pricing

Custom pricing available upon contacting sales

Fireworks AI Specifications

Fireworks AI is available on web app. It offers an API and has a free trial.

Deployment
  • Cloud Based
Billing cycle
Monthly
Desktop
  • Web App
Languages
  • English
Built for
  • Startups
  • SMEs
  • Enterprises
Support
  • Email
  • Tickets
Integrations
OpenAI-compatible API, Microsoft Foundry/Azure, LangChain
Public API
Yes
Free trial
Yes
Free plan
No
Runs in browser
Yes
Customisable
Yes
Website
fireworks.ai

Fireworks AI Videos

Fireworks AI Screenshots

Fireworks AI screenshot 1

Used Fireworks AI? Write the first review.

Fireworks AI FAQs

Yes, new accounts receive $1 in free credits to try serverless inference, roughly enough for around 1M tokens on a 70B-class model.

Fireworks AI offers serverless access to more than 400 open source models spanning language, vision, image, and audio.

Yes, it supports LoRA and full-parameter supervised fine-tuning, DPO, and reinforcement fine-tuning, priced per million training tokens.

Dedicated deployments run on NVIDIA H100, H200, B200, and B300 GPUs, billed per GPU-hour.

Yes, it is SOC 2 Type II, GDPR, and HIPAA compliant, with private deployment and audit log options for enterprise customers.