Skip to main content

API changelog

Changes to the public API, newest first. Breaking changes are announced here before they ship, as set out in the versioning policy.

2026-11-13​

Docs

  • New pages: Errors, MCP server, this changelog and the versioning policy.
  • Base URL corrected to https://api.funai.art in every example. Earlier examples used hosts that do not exist.
  • Examples now use real model IDs (gpt-4o-mini, gpt-image-1, flux-schnell, veo3-1-fast, elevenlabs-tts-multilingual-v2).
  • Native generation paths corrected to /generation/v1/generate… and /generation/v1/jobs/{job_id}/status.
  • Authentication documents both accepted headers: Authorization: Bearer and X-API-Key. API keys are created on developers.funai.art.
  • Pricing & balance: API calls are paid from a separate USD API balance on the developer platform; credits convert at one rate, 10,000 credits = $1.
  • OpenAI-compatible base URL is now https://api.funai.art/v1 (the old /generation/v1 base still works). GET /v1/models is public.
  • API reference rendered from the public OpenAPI spec (/openapi.json), including /v1/models, /v1/images/* and /v1/videos*.
  • POST /v1/images/generations and POST /v1/videos honour Idempotency-Key; a slow image answers 202 with a job and poll_url instead of timing out.
  • Errors: 25 stable error types; /v1 endpoints use the OpenAI error shape.
  • Webhooks: the events actually delivered (generated from the platform's webhook routing), the X-FunAI-Signature header (timestamped HMAC), the payload shape, and per-request callback_url.
  • Rate limits: rate-limit headers are sent on 429 responses.