Developers
Ship in minutes, not sprints
Everything you need to send your first request, migrate an existing client, and run it in production.
Your first response in minutes
Make your first inference call in minutes: create an API key, POST a prompt to /api/ainf/inference with a bearer token, and read tokens, latency and price back.
Explore →One endpoint, fully documented
Reference for POST /api/ainf/inference: bearer auth, the Idempotency-Key header, request fields, the result object and every error status you can expect.
Explore →Switch over in an afternoon
Move a chat-style integration to AI Inference: change the base URL, keep your bearer header, map your message text to prompt, and keep your parameters.
Explore →Limits that protect you, not slow you down
How rate limits work: a per-account requests-per-minute ceiling set in settings, 429 responses with a retry hint, and cache hits that skip the limit.
Explore →Retry freely, get billed once
Send an Idempotency-Key header on POST /api/ainf/inference so a retried request is metered once. Keys are namespaced to your account and safe to retry.
Explore →