OpenAI rebuilds ChatGPT Voice with full-duplex speech models and GPT-5.5
The Batch (Andrew Ng) · rss · 2026-07-22
OpenAI has rebuilt ChatGPT Voice around a new pair of speech models, GPT-Live-1 and GPT-Live-1 mini, which process audio continuously instead of turn by turn.
- The system is full-duplex: it can listen and speak at the same time, while routing harder requests to GPT-5.5 in the background.
- The app now supports live translation, remastered preset voices, and selectable reasoning effort for GPT-Live-1.
- OpenAI says GPT-Live-1 scored 84.2% on GPQA at high reasoning, versus 45.3% for the older Advanced Voice Mode.
- In browsing tasks, GPT-Live-1 reached 75.2% on BrowseComp, compared with 0.7% for AVM.
- The feature is live globally on iOS, Android, and ChatGPT.com; it is included in Go, Plus, Pro, and the free tier, but there is no developer API yet.
- OpenAI also appears to be setting up voice as a first-class interface for future devices, including a rumored screenless smart speaker.
More from Models
- Same Echo Maze prompt, three frontier models: all passed visually but shipped the same hidden bug — eyishazyer · 2026-09-11
- Benchmark scores drop from 89% to 19% on new evals — how benchmaxxing breaks leaderboard trust — airesearch12 · 2026-09-11
- ChatGPT tells user their question is too hard and to 'accept dumber answers' — phido3000 · 2026-09-11
- Developer Building a Unified Leaderboard of All Model Benchmark Scores — airesearch12 · 2026-09-11
- Rumor claims Kimi faked performance by serving Claude; DeepSeek new model surprises in evals — realsohamparekh · 2026-09-11
- GPT-5.6 writes well but is instantly forgettable, user complains — BasedRaddka · 2026-09-11