MiniCPM-V 4.7 (35B-A3B, Q4)
OpenBMB's vision-language model with 35B total parameters and roughly 3B active per token. Understands images, documents, and multiple images, with native tool calling and optional thinking. The Q4_K_M download includes the 16x vision projector. Requires llama.cpp with MiniCPM-V 4.7 support (upstream PR #29416); Gezel's v0.6.0 engine cannot run its vision encoder.
| At a glance | |
|---|---|
| Made by | OpenBMB |
| Download size | 22.3 GB |
| Parameters | 35B |
| Quantization | Q4_K_M |
| Context window | 262K tokens |
| Uses tools | Yes |
| Best for | Everyday work |
| Runs on | llama.cpp, MLX |
| License | Apache 2.0 |
| Added to the catalog | 2026-10-13 |
Gezel downloads this model the first time a gezel needs it, checks the file against its published hash, and applies tuned settings for it.