openbmb/MiniCPM-V-4.7-35B-A3B
main boardmodel
MiniCPM-V 4.7 is the latest model in the MiniCPM-V series, available in a 35B-A3B variant, with 35B-A3B being the first MoE model in the MiniCPM-V series. Mixture-of-experts, 35B total, 3B active (256 experts, 8 per token). Context: 256K tokens (config). License: apache-2.0.
- Org
- openbmb
- Released
- 2026-10-07
- Status
- main board
- License
- apache-2.0 (permissive)
- Parameters
- 35.2B total, 3.0B active
- Architecture
- MoE · 256 experts, 8/token · MiniCPMV4_7ForConditionalGeneration
- Context
- 262,144 tokens
- Task
- image-text-to-text · vision
- Signal
- 1088 HF likes · 218995 downloads · 1 X post(s)
Runs on
real quant files; need = weights + FP16 KV cache for 8K tokens + 1GB. A sizing calculation, not tested.
- Q4 ~22.9 GB: 24GB GPU (3090 / 4090), 32GB GPU (5090), 24GB GPU (7900 XTX), Mac 36GB, Mac 48GB, CPU only, 32GB RAM (slow)
- Q8 ~38.6 GB: 48GB GPU (RTX 6000 Ada / A6000), Mac 64GB (M1 Max), CPU only, 64GB RAM (slow)
- FP16 ~71.9 GB: 96GB GPU (RTX Pro 6000 Blackwell), 1× DGX Spark (128GB), 2× DGX Spark (256GB), Strix Halo 128GB unified (~96GB usable), Mac 96GB, Mac 128GB, Mac 192GB, Mac 256GB, Mac 512GB, CPU only, 128GB RAM (slow), CPU only, 256GB RAM (slow)
Links
- Hugging Face
- Ollama
- GGUF: bartowski/MiniCPM-V-4.7-35B-A3B-GGUF
- MLX: mlx-community/MiniCPM-V-4.7-35B-A3B-4bit
- MLX: mlx-community/MiniCPM-V-4.7-35B-A3B-5bit
- MLX: mlx-community/MiniCPM-V-4.7-35B-A3B-6bit
- MLX: mlx-community/MiniCPM-V-4.7-35B-A3B-8bit
- MLX: mlx-community/MiniCPM-V-4.7-35B-A3B-OptiQ-4bit
- MLX: mlx-community/MiniCPM-V-4.7-35B-A3B-mxfp4
- MLX: mlx-community/MiniCPM-V-4.7-35B-A3B-mxfp8
- MLX: mlx-community/MiniCPM-V-4.7-35B-A3B-nvfp4
- quant: openbmb/MiniCPM-V-4.7-35B-A3B-AWQ
- quant: openbmb/MiniCPM-V-4.7-35B-A3B-FP8
- quant: openbmb/MiniCPM-V-4.7-35B-A3B-GPTQ
First published here 2026-10-07 · updated 2026-10-07 · markdown · JSON