Alibaba launches Qwen-Audio-3.1-Realtime, cutting background-speech false responses from 73% to 13%
mark_k · x · 2026-09-30
Alibaba released Qwen-Audio-3.1-Realtime, a realtime voice model available via API.
- Full-duplex: listens while speaking, calls tools mid-conversation
- Decides when to speak, stop, or stay silent
- Background-speech false responses dropped from 73% to 13% vs the previous model
The pitch: a voice agent that follows evolving requests and knows when not to talk.
More from Models
- JevBench details revealed: private question pools and sealed test sets to stop benchmark gaming — airesearch12 · 2026-09-30
- OpenAI's Dots device rumored to require a Pro 200 subscription plan — testingcatalog · 2026-09-30
- Longtime User Reports Gemini Has Gotten Worse: Reminders and Samsung Notes Integrations Broke — Octane2100 · 2026-09-30
- VLM Chain-of-Thought Doesn't Reliably Track Visual Evidence, EMNLP Paper Finds — oanacamb · 2026-09-30
- Six frontier models benchmarked across 34 capabilities in nine computer vision areas — ducha_aiki · 2026-09-30
- Anthropic Ships Sonnet 5.5: Real-World Test on Website Build and Multi-Currency Sheet — Rasmic · 2026-09-30