Deepgram Nova-3
Deepgram's current-generation streaming + batch ASR model.
Deepgram's current-generation streaming + batch ASR model.
Best for real-time + batch transcription where Deepgram's latency + cost are the differentiator. Pricing: from $0.0043/min (Nova-3, batch).
What it is
Nova-3 is Deepgram's current-generation ASR model exposed under model=nova-3 on the existing /v1/listen API. Real-time streaming and batch are both supported, with diarization, word timestamps, smart formatting and language detection. Distinct from the high-level 'deepgram' entry already in this directory; covered separately here to make model selection visible at the directory level. Best fit: real-time + batch transcription where deepgram's latency + cost are the differentiator. Caveats: distinct from the 'deepgram' entry — same vendor, model-specific sku. Pricing as listed: from $0.0043/min (Nova-3, batch). Feature flags from vendor docs: speaker diarization, word-level timestamps, streaming, HIPAA-eligible under BAA. Directory tags: commercial-api. Last vendor-page check: 2026-05-12.
Watch out for: Distinct from the 'deepgram' entry — same vendor, model-specific SKU.
Install / use
Features
| Speaker diarization | Yes |
| Word-level timestamps | Yes |
| Streaming / real-time | Yes |
| Languages supported | None |
| HIPAA eligible | Yes |
Deepgram Nova-3 vs Whipscribe
| Feature | Deepgram Nova-3 | Whipscribe |
|---|---|---|
| Category | Transcription APIs | Transcription APIs |
| Pricing | from $0.0043/min (Nova-3, batch) | free beta |
| Speaker diarization | Yes | Yes |
| Word timestamps | Yes | Yes |
| Streaming | Yes | No |
| Languages | — | 99 |
| Platforms | API | Web, API, MCP |
Alternatives to Deepgram Nova-3
Whipscribe is a managed faster-whisper + whisperX service. If you want transcripts without running infrastructure, paste a URL or drop a file in the form below — you'll have a transcript in seconds.