Tarseek
Sign in

Tarseek API

Platform coverage

Tarseek is an early-stage product. This page summarizes what the public API supports today and what is planned. It is not a commitment that planned items will ship on a particular date.

Available now

  • Bearer API keys created in the dashboard
  • OpenAI-compatible chat completions, including streaming
  • Model catalog with developers, deployments, pricing, and capability metadata
  • Same-model failover, circuit breaking, price ceilings, and allow/block lists
  • Vision in chat on models that accept image input
  • Tool calling, structured output, and reasoning_effort where the model supports them
  • Normalized errors, request IDs, prepaid credit reservation, and usage/cost on successful chat responses
  • Project rate limits (RPM, TPM, concurrency) with 429 responses and retry headers

Planned

  • Dedicated embeddings, image generation/editing, and audio APIs
  • Public retrieve-by-id on GET /v1/models/{id}
  • Request-level provider preference and latency-based routing
  • Key expiry, distinct revoked-key errors, and optional key scopes
  • Self-serve checkout for purchasing credits
  • Public usage endpoint and dated API versioning

See Models and the API reference for endpoint details.