MiniCPM-o Omnimodal Series
MiniCPM-o is an omnimodal model series supporting vision, speech and text. It can see, listen and speak at the same time, and supports full-duplex real-time interaction.
Series Overview
| Model | Parameters | Architecture | VRAM | Highlights | HuggingFace |
|---|---|---|---|---|---|
| MiniCPM-o 4.5 | 9B | SigLIP2 + Whisper + CosyVoice2 + Qwen3-8B | 19 GB | Full-duplex real-time interaction, approaching Gemini 2.5 Flash | Download |
| MiniCPM-o 2.6 | 8B | SigLIP + Whisper + ChatTTS + Qwen2.5-7B | — | Matches GPT-4o-202405 | Download |
Related Resources
| Topic | Link |
|---|---|
| Code repository and release notes | OpenBMB/MiniCPM-o |
| MiniCPM-o 4.5 model weights | Hugging Face / ModelScope |
| MiniCPM-o 2.6 model weights | Hugging Face / ModelScope |
| Model documentation | MiniCPM-o 4.5 / MiniCPM-o 2.6 |
| Online demo | MiniCPM-o 4.5 live demo |
| Local real-time demo | MiniCPM-o-Demo |
| API access | API guide |
| Technical report | MiniCPM-o 4.5 Technical Report |
Developer Showcase
View all →Booking Desk: Full-Duplex Voice BookingA no-fine-tuning full-duplex appointment desk driven by a state machine and judgeGander: Talk While It WorksFull-duplex omni agent that delegates long tasks to Codex mid-conversationI Had an Agent Fine-tune a Multimodal Model from Scratch68 minutes later it handed over a plan that just runsTyloQuant: Mixed-Precision QuantizationMFQ mixed-precision quantization that compresses MiniCPM-o 4.5Grandpa's Bedtime Stories: Photo TalesA WebXR storybook where Grandpa answers voice questions about photosAI Jarvis: Screen-Aware Desktop PetA local Windows desktop assistant that watches your screen and audioYRobot: Full-Duplex Reachy MiniFull-duplex audio-visual conversation and lifelike motion for Reachy Mini