Capabilities
What is available today and what is coming. The same list is returned by GET /v1/models.
| Capability | Status | Request |
|---|---|---|
| Language set in the request | Available | language.explicit |
| Automatic language detection | Coming soon | language.auto |
| Native mode | Available | mode.native |
| Code-mixed mode | Available | mode.codemix |
| Verbatim mode | Available | mode.verbatim |
| Segment timestamps Each segment carries its start and end time; in the 22 Indian languages segments are sentence-sized. | Available | timestamps.segment |
| Word timestamps | Coming soon | timestamps.word |
| Speaker labels Speaker labels on segments; works best with two to five speakers. | Available | diarization |
| Mixed channels | Available | channels.mix |
| Separate channels (stereo calls) For stereo recordings such as call-centre calls: each channel transcribed on its own, segments labelled by channel. | Available | channels.separate |
| WAV | Available | input.wav |
| FLAC | Available | input.flac |
| MP3 | Available | input.mp3 |
| M4A / AAC | Available | input.m4a |
| OGG / Opus | Available | input.ogg |
| WebM audio | Available | input.webm |
| Video files | Coming soon | input.video |
| JSON result | Available | output.json |
| Plain text (the result’s text field) | Available | output.txt |
| SRT subtitles | Coming soon | output.srt |
| WebVTT subtitles | Coming soon | output.vtt |
How coming-soon works
A coming-soon capability is part of the published roadmap but not yet part of the supported contract. Requests for it are rejected with 422 capability_unavailable and a detail naming it, so your integration fails loudly rather than getting something approximate. When it launches, it appears here, in GET /v1/models and in the changelog.