v26.9.7 livekit-agents@1.8.1
CaviraOSS/LongMemorySep 9, 2026 → Sep 10, 20261 day between releases
What's New
Introduces full-duplex speech model support via DuplexModel class and OpenAI GPT-Live.
Breaking Changes
- New DuplexModel class requires code updates to utilize full-duplex capabilities
- Speechmatics V2 migration may require updates to existing STT configurations
New Features
- DuplexModel class for simultaneous speaking and listening
- OpenAI GPT-Live model support
- Phonic plugin watermarking configuration
- Speechmatics V2 migration
Improvements
- Connection pooling and TTS stability improvements
- Isolated ruff dependencies in CI
- Telemetry option object descriptions in session report
Bug Fixes
- Fixed TTS prewarm replacement logic
- Drained unused adapter metrics streams
- Dropped unused audio forwarding buffer
Upgrade Notes
Update code to use the new DuplexModel class for simultaneous listening and speaking. Review Speechmatics V2 migration requirements and update Phonic plugin watermarking configuration as needed.
The Two Releases
v26.9.7
Sep 9, 2026· 1 month ago
The v26.9.7 release focuses on fixing voice discovery in Docker, notebook launcher issues, and cleaning up version history. It includes crucial update instructions regarding folder deletion.
Full release noteslivekit-agents@1.8.1
Sep 10, 2026· 1 month ago
A major feature release introducing full-duplex speech model support via the new DuplexModel class, featuring OpenAI's GPT-Live, along with various connection and plugin stability improvements.
Full release notes