Plain-English Summary
A live transcription model designed for voice agents, meetings, and conversational applications.
Scribe v2 Realtime
A live transcription model designed for voice agents, meetings, and conversational applications.
Scribe
Audio / Speech
GA
Open weights: No
API: Yes
Technical Notes
Streaming speech recognition model with approximately 150ms end-to-end latency and multilingual real-time output.
Best For
Voice agents, live captions, meeting assistants, real-time translation
Watch Out For
Streaming integrations are more complex and may trade some batch accuracy for latency.
Model Specs
- Maker
- ElevenLabs
- Country
- USA
- Family
- Scribe
- Type
- Audio / Speech
- Release date
- 2025-11-11
- Date confidence
- Exact
- Status
- GA
- Context window
- Varies
- Max output
- Not listed
- Parameters
- Not disclosed
- Active parameters
- Not disclosed
- Architecture
- Proprietary streaming ASR model
- Reasoning
- Not applicable
- Tool calling
- No
- Structured output
- Streaming transcript events
- API available
- Yes
- Open weights
- No
- Self-hostable
- No
- License
- Proprietary
- Inputs
- Streaming audio
- Outputs
- Streaming text
- Approx. price
- $0.39 per audio hour
- How to access
- Provider website or API
- Model / API ID
- scribe_v2_realtime
- Knowledge cutoff
- Not publicly disclosed
- Fine-tuning
- Key terms and streaming settings
- Completeness
- 81 %
- Verified
- Jul 18, 2026
- Snapshot date
- Jul 18, 2026
- Source
- Official source