ByteDanceSeedRealtime
SeedRealtime
SeedRealtime unifies text, audio and visual inputs for live conversation. ByteDance announced a full product rollout on 5 August 2026.
Technical guide
The model can respond to a scene change without a new typed request. The release also describes turn-taking, interruption handling and tool calls in continuous conversations.SeedRealtime release
Practical use
Practical guidance. Test interaction timing separately from answer quality. A useful assistant should respond to the intended speaker and scene change without repeatedly interrupting ordinary activity.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Provider product |
| Reviewed access | Product rollout is confirmed; this guide does not establish a public standalone API.SeedRealtime release |
Related models
- Seed LiveInterpret 2.0
ByteDance ยท Seed LiveInterpret
End-to-end speech interpretation model for Chinese and English.
Launch date API