OpenBMB's open 9B MiniCPM-o 4.5 sees, listens and speaks simultaneously via Omni-Flow
kimmonismus · x · 2026-10-23
- The author highlights China's concrete work in real-time full-duplex: OpenBMB's MiniCPM-o 4.5, a 9B-parameter open model that can see, listen and speak simultaneously.
- The key technique is the Omni-Flow framework, which aligns incoming and outgoing streams on a shared timeline.
- The paper argues the bottleneck is no longer modality coverage or latency but the interaction paradigm itself: perception and response remain in alternating phases and models stay reactive. MiniCPM-o 4.5 also shows proactive behaviors — reminders or comments based on continuous scene understanding.
Related event: Sequoia China and Hillhouse Back Full-Duplex Startup Telos(2 posts)→
More from Models
- Models: useless or misleading for extrapolation, yet rock-solid on known-knowns — teodorio · 2026-10-23
- Diffusion LLM Mercury 3 beats Claude Sonnet 5.5 and GPT-6 Luna on simulated air traffic control — volokuleshov · 2026-10-23
- ThinkingCap squeezes 38.5% fewer thinking tokens out of Qwen3.8-Flash-Next at ~1pp accuracy cost — Thom_Wolf · 2026-10-23
- DeepSeek ships open-weight V4.1-Flash-1017, Terminal-Bench jumps 31.2 to 42.3 — petrusenko_max · 2026-10-23
- DeepSeek ships V4.1-Flash-1017: Terminal-Bench jumps 31.2→42.3, MIT-licensed — petrusenko_max · 2026-10-23
- OpenAI's Decisions API hits GA: near real-time routing up to 10x faster than GPT-6 Luna — OpenAIDevs · 2026-10-23