Provider pricing

Anyscale AI API pricing

Anyscale is the company behind Ray, and its Endpoints product serves open-weight models — Llama 4, Mistral, Qwen3 — on optimized Ray Serve clusters. The same platform handles distributed training, RAG pipelines, and custom batch jobs, making it a natural fit for teams that want LLM serving plus the broader Ray ecosystem.

Founded2019
HeadquartersSan Francisco, CA
Tracked modelsDirectory

Best for

Teams already using Ray, or those needing inference + batch + training on one platform.

Strengths

  • Ray Serve under the hood — proven at internet scale
  • Endpoints for inference, Jobs for batch, Workspaces for development
  • Bring-your-own-cloud option (AWS, GCP, Azure) for data residency
  • Strong fine-tuning and RAG tooling on the same platform