Choose plan

Unlock everything the Kestrel Foundry API can do.

Prototype

Free to start. Requests may be sampled to improve our systems.

  • Call every general-availability model
  • Build and deploy agents
  • Ship a first prototype in an afternoon
Free
Start prototyping

Scale

Pay for what you run. Billed monthly or drawn from prepaid credits.

  • Metered by the token; see the rate card
  • Access to research-preview models
  • Hosted agent runtime with autoscaling workers
  • Access to the fine-tuning API
  • Max. 8 requests per second, raised on request
  • Max. 3M input tokens per minute, raised on request
  • Max. 12B input tokens per month, raised on request
  • 24M tokens per minute, and 240M per month, for Kestrel Embed
  • Every feature at launch, betas and previews included
  • Requests are never used to improve our systems
Price Metered
Pay as you scale