A face wall smart open source MiniCPM-o 4.5,9B parameters for full-time dialogue

The All-Script Flag Model MiniCPM-o 4.5, with only 9B parameters, achieves a "side-vision, side-hearing, pro-active" dual-work capability and becomes the industry's first “free-in-time” model; the model uses an end-to-end full-modular structure, using a time-sharing and active interaction mechanism, to automatically decide whether to speak at a 1Hz frequency, and to achieve perception and sound-based dialogue; and a comprehensive upgrading of voice generation, to support voice cloning and role-playing, to stabilize long-speech synthesis, and to provide an open source llama.cpp-omni framework for high-impact side reasoning。

Search