Changelog

Changes to the API, SDKs and console. Additive changes ship without notice; anything that could break an integration is announced first.

2026-10-31

API address: projects in India are served at https://api.scriba.ai (shown under API keys and Settings). The previous address keeps working.
Status: each region answers GET /status with separate API and transcription availability; the status page shows both.
Short audio: when the synchronous endpoint is already handling as many requests as it takes at once, extra requests get 503 capacity_unavailable with Retry-After immediately. Nothing is charged.
Deletion: deleted audio stays deleted, including anything uploaded afterwards through a link issued before the deletion.
Console: a retry after a lost reply never starts a second transcription — even after a reload (when the browser allows site storage), the playground finds the one that was started; large uploads continue where they stopped after a failure, a cancel or a reload; the transcript shows every warning and marks stretches that could not be transcribed; clearer error messages; works on phones; password-reset links work on any device.

Transcriptions report progress (percentage of the audio transcribed).
GET …/result?format=txt returns the plain-text transcript.
Webhook endpoints are verified with a signed event when they are registered; new calls to list, verify, rotate the signing secret (24-hour overlap), disable, read delivery history and replay a delivery.
Console: replace an API key in one step; verify, rotate and resend webhooks; export usage as CSV.
Audit log for owners and admins (console → Account → Audit log), with CSV export.
Automatic deletion is on: audio 24 hours after a transcription finishes and transcripts after 30 days by default (adjustable per project); request logs 90 days, audit log 1 year.

Console: sign-up, organizations and projects with a region per project, API keys, playground, transcriptions, usage and credit, request logs, webhooks with delivery history, retention settings, team roles.
API: asynchronous transcriptions with direct and resumable uploads; synchronous short audio; 22 Indian languages and English; native, code-mixed and verbatim modes; speaker labels; separate channels; segment timestamps; signed webhooks; usage and credit.
SDKs: Python and TypeScript.
Coming soon: automatic language detection, word timestamps, SRT and WebVTT output, video files.

The API path carries its major version (/v1). Within a version we only add: new endpoints, new optional fields, new enum values, new warning codes. Write clients that ignore fields they do not know.
A change that could break an integration goes into a new version. The old version keeps working for at least six months after the new one is announced here and by email to organization owners.
Results report the model and version that produced them.