Skip to main content
Use price search when you need live rates before you run a workload or compare hosts. The endpoint is public. No API key required.

Search by model name

Pass a model ID or partial name as q:
Each row in data includes:
  • model_id
  • provider_id
  • pricing.prompt and pricing.completion in USD per token

Read the rates

Multiply by 1_000_000 when you want dollars per million tokens: Filter rows where model_id matches exactly when the query returns variants or aliases.

Paginate large result sets

Add page and page_size when a model appears on many hosts:
Stop when meta.page reaches meta.total_pages.

Next step

Send token counts to POST /v1/traces/cost to get a dollar total for one request.