AssemblyAI-STT-3.7-Realtime vs Scribe v3 Live
Compare documented capabilities and access. A dash means a comparable figure is unavailable.
| Specification | ||
|---|---|---|
| Task | Speech recognition | Speech recognition |
| Published capabilities | Live conversations. 32 languages in one stream with automatic detection. | Transcribes live audio with partial results and turn prediction. ElevenLabs reports 43% fewer errors than Scribe v2 Realtime in its launch evaluation. |
| Access | Realtime Speech-to-Text API as assemblyai-stt-3-7-realtime | ElevenLabs Speech to Text API |
| Result | Streaming transcripts with speaker labels at turn boundaries | Streaming partial and committed transcripts |
| Billing terms | $0.30 USD per hour of audio on AssemblyAI. | Not published |
| Sources | Publisher documentation · Additional documentation | Publisher documentation · Additional documentation |
| Inputs | Audio | Audio |
| Weights | Weights unconfirmed | Weights unconfirmed |
| Released | Nov 18, 2026 | Nov 9, 2026 |