ElevenLabs Scribe Models by ElevenLabs
ElevenLabsProprietary
3 models2025–2026
Last refreshed 2026-06-29. Next refresh: weekly.
Details
ResearcherElevenLabs
LicenseProprietary
Commercial useCommercial use: conditional
Models3
Released2025–2026
Links
WebsiteAbout
ElevenLabs' speech-to-text (Scribe) model family. Scribe v2 (January 2026) supports 90+ languages, 32-speaker diarization, 56 entity types, and keyterm prompting. Scribe v2 Realtime (~150ms) is optimized for voice agents. Scribe v1 is deprecated — migrate to v2.
Decision facts
- Best fit
- speech recognitionagent workflows
- Capability starting point
- Scribe v2
- Lowest tracked input
- Not tracked
- Closest related family
- ElevenLabs Text-to-Speech
Current Variants
Use-when guidance is based on each model's tracked capabilities, context window, release date, and replacement status.
2 in view1 retired
Scribe v2Current
Use when the workload needs speech recognition and audio.
2026-01speech recognitionaudio
Scribe v2 RealtimeCurrent
Use when the workload needs speech recognition and audio.
2025-11speech recognitionaudio
| Model | Use when | Released | Signals | Status |
|---|---|---|---|---|
| Scribe v2 | Use when the workload needs speech recognition and audio. | 2026-01 | speech recognitionaudio | Current |
| Scribe v2 Realtime | Use when the workload needs speech recognition and audio. | 2025-11 | speech recognitionaudio | Current |
Release Timeline
3 release groups2026-01
1 current
Scribe v2
Currentspeech recognitionaudio
2025-11
1 current
Scribe v2 Realtime
Currentspeech recognitionaudio
2025-02
1 retired
Scribe v1
Archivedspeech recognitionaudio
Specifications(3 models)
| Model | Released |
|---|---|
| Scribe v2 | 2026-01 |
| Scribe v2 Realtime | 2025-11 |
