Deploy and run LLMs at lightning speed with Fireworks AI. Optimized for performance and scalability.
About
Fireworks AI is a high-performance inference platform built specifically for generative AI, designed to provide developers and enterprises with lightning-fast model deployment and inference services. The platform integrates the latest open-source large language models (LLMs) and image generation models, supporting a wide range of tasks from text understanding to visual creation. Whether you are calling pre-hosted models directly or uploading your own models for fine-tuning and deployment, Fireworks AI helps you bring AI capabilities into production quickly with industry-leading speed and reliability.
Key Features
Pricing & Fees
- Pricing: Serverless inference is pay-per-token with pre-paid credits (e.g. Kimi K3 and Ember-1 at $3/M input and $15/M output). Managed training starts at $0.50 per 1M training tokens for LoRA SFT on models up to 16B, scaling to $40 for Full Param DPO on models over 300B.
- Embeddings: Priced by base model parameter count — from $0.008 per 1M input tokens for models up to 150M parameters, up to $0.1 per 1M for Qwen3 8B (limited-time offer).
- Free start: Muse Glimmer 30B SFT is free via UI and the Serverless Training API until 11/30, with no credit card required.
- Billing details: Usage is billed on output, input, and cached tokens; Auto Reload tops up automatically when your balance runs low, and you can set a monthly spend limit. Fine-tuned models are served at the same price as base models.
- Deployment options: Serverless (pay per token with Priority and Fast options), On-Demand (pay per GPU second, dedicated deployments), and Reserved (guaranteed capacity and higher quotas); APIs are OpenAI and Anthropic compatible.
Who It's For
AI application developers who need to quickly integrate text or image generation capabilities into chatbots, content creation tools, and design assistant apps. Machine learning engineers who want to fine-tune pretrained models to produce domain-specific models faster without the cost and time of training from scratch. Startup teams and independent developers with limited budgets who need cost-effective AI inference services — Fireworks AI's free tier and pay-as-you-go pricing help keep initial costs under control. Enterprise IT departments seeking to embed generative AI capabilities securely and reliably into internal systems such as intelligent customer support and automated report generation.
