Developers / Changelog

Changelog.

Highlights from each API release. Breaking changes, deprecations and removals are listed with the release that introduces them.

Current version
API v1 · Stable

All /api/v1/ endpoints are active and not deprecated.

Versioning
Major in URL, additive in place

Breaking changes increment the major version in the URL path. Additive changes ship without a version bump.

v1.2.0 API release

September 24, 2026.

  • Discover the job types the API accepts with GET /api/v1/task-types.
  • Job status now covers the full lifecycle, including ingesting, failed_partial and cancelled.
  • Rate-limit errors say which limit was hit and when to retry.
  • OpenAI-compatible transcription is available as a pilot, with its limits listed in the API reference.
  • Webhook deliveries survive coordinator restarts and are protected against DNS rebinding.
v1.1.0 API release

June 10, 2026.

  • List available models with GET /api/v1/models, and choose a backend, model or quality per job.
  • A second, faster English model (Parakeet), used automatically for English jobs when available.
  • Set language to route non-English audio to the right model.
  • Submit many jobs at once with POST /api/v1/jobs/batch.
  • Jobs are accepted right away (202) while audio is prepared; preparation errors show on job status.
  • API errors include a request ID and a machine-readable code.
v1.0.0 API release

April 19, 2026.

  • First stable release of the /api/v1/ transcription API: submit audio, track progress, fetch the transcript.
  • Authenticate with your API key as a bearer token.