Fireworks AI
Inference
Official site consultedAI and speech (global tools)
Last verified: 1 October 2026
Fireworks AI offers serverless inference billed per token, dedicated GPU deployments and model training. Over 100 models cover text, vision, audio, image and embeddings, with function calling, structured JSON outputs and batch inference. The API is OpenAI-compatible.
Services
- Serverless per-token inference
- Dedicated GPU deployments with autoscaling
- Training and fine-tuning
- Function calling and structured JSON outputs
- Embeddings with reranking, batch inference
Access requirements
The page gives no sign-up link, pricing or free credits.
Sources
- Fireworks AI (new tab), verified on 1 October 2026Official site consulted