MiniCPM V 4.7 35B A3B needs about 23.5 GB of memory at Q4_K_M (a 22.3 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 26 quants, from 0.4 GB to 69.4 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 2.7 GB.
MiniCPM‑V 4.7 35B‑A3B – a multimodal MoE model with on‑device vision support
MiniCPM‑V 4.7 35B‑A3B is a 34.7 B parameter model that uses a mixture of experts. It supports text and image input and can run on mobile and local devices. The model is released under the Apache‑2.0 license.
Good at
- Vision‑enabled multimodal inference
- Efficient on‑device performance
- Large context length of 262144 tokens
Limits
- Requires a 35 B‑A3B MoE architecture
- No speculative decoding support
- Limited to text and image modalities
Best for
- Mobile vision‑assistant apps
- Local multimodal research prototypes
- Edge inference with limited GPU memory
Who it’s for. Developers building on‑device multimodal applications who need a balance of vision capability and efficient inference. The model is suitable for those who can deploy it via llama.cpp, SGLang, or other supported frameworks.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| — | 0.4 GB | 1.6 GB | Fits | Get file → | |
| IQ2_XXS | 10.8 GB | 12.0 GB | Too big | Get file → | |
| IQ2_XS | 11.3 GB | 12.4 GB | Too big | Get file → | |
| IQ2_S | 12.0 GB | 13.1 GB | Too big | Get file → | |
| IQ2_M | 13.2 GB | 14.4 GB | Too big | Get file → | |
| Q2_K | 13.8 GB | 15.0 GB | Too big | Get file → | |
| IQ3_XXS | 15.6 GB | 16.8 GB | Too big | Get file → | |
| IQ3_XS | 16.3 GB | 17.5 GB | Too big | Get file → | |
| Q3_K_S | 16.3 GB | 17.5 GB | Too big | Get file → | |
| Q3_K_M | 17.3 GB | 18.5 GB | Too big | Get file → | |
| Q3_K_L | 18.3 GB | 19.4 GB | Too big | Get file → | |
| IQ3_M | 19.0 GB | 20.1 GB | Too big | Get file → | |
| IQ4_XS | 19.9 GB | 21.1 GB | Too big | Get file → | |
| Q4_0 | 19.9 GB | 21.1 GB | Too big | Get file → | |
| Q4_K_S | 20.9 GB | 22.1 GB | Too big | Get file → | |
| Q4_1 | 22.0 GB | 23.1 GB | Too big | Get file → | |
| IQ4_NL | 22.3 GB | 23.5 GB | Too big | Get file → | |
| Q4_K_M | 22.3 GB | 23.5 GB | Too big | Get file → | |
| Q4_K_L | 24.1 GB | 25.3 GB | Too big | Get file → | |
| Q5_K_S | 25.1 GB | 26.3 GB | Too big | Get file → | |
| Q5_K_M | 27.0 GB | 28.2 GB | Too big | Get file → | |
| Q6_K_S | 29.4 GB | 30.5 GB | Too big | Get file → | |
| Q6_K | 30.7 GB | 31.9 GB | Too big | Get file → | |
| Q6_K_L | 32.3 GB | 33.5 GB | Too big | Get file → | |
| Q8_0 | 36.9 GB | 38.1 GB | Too big | Get file → | |
| BF16 | 69.4 GB (2 parts) | 70.5 GB | Too big | Get file → |
For 8 GB VRAM use the Q4_K_M quant (22.32 GB file, ~23.5 GB memory at 8 k context). For 12 GB VRAM the IQ4_XS quant (19.89 GB file, ~21.1 GB memory) is a good trade‑off. The BF16 variant (69.38 GB file) requires at least 70.5 GB memory and is best for high‑end servers.
What GPU do you need for MiniCPM V 4.7 35B A3B?
These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.
- 4 GB VRAM: the best quant that fits is the main file (0.4 GB file, about 1.6 GB in use).
- 6 GB VRAM: the best quant that fits is the main file (0.4 GB file, about 1.6 GB in use).
- 8 GB VRAM: the best quant that fits is the main file (0.4 GB file, about 1.6 GB in use).
- 12 GB VRAM: the best quant that fits is IQ2_XXS (10.8 GB file, about 12.0 GB in use).
- 16 GB VRAM: the best quant that fits is Q2_K (13.8 GB file, about 15.0 GB in use).
- 24 GB VRAM: the best quant that fits is IQ4_NL (22.3 GB file, about 23.5 GB in use).
- 32 GB VRAM: the best quant that fits is Q6_K (30.7 GB file, about 31.9 GB in use).
- 48 GB VRAM: the best quant that fits is Q8_0 (36.9 GB file, about 38.1 GB in use).
Which MiniCPM V 4.7 35B A3B quant should you choose?
Q4_K_M is the usual starting point for MiniCPM V 4.7 35B A3B: it balances file size and quality well. Here it is a 22.3 GB file.
The Q8_0 file is 36.9 GB, about 1.7 times the size of Q4_K_M. It keeps the weights closer to the original, which only matters if you have the memory to spare.
If memory is tight, the smallest file is the first one in the table at 0.4 GB, with the quality trade-offs that come with fewer bits.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about MiniCPM V 4.7 35B A3B
How much VRAM does MiniCPM V 4.7 35B A3B need?
About 23.5 GB at Q4_K_M with an 8K context window. That is the 22.3 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.
Can I run MiniCPM V 4.7 35B A3B on an 8 GB GPU?
Yes, with the smaller files: the 0.4 GB file needs about 1.6 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of MiniCPM V 4.7 35B A3B GGUF?
It depends on the quant: file is 0.4 GB, IQ2_XXS is 10.8 GB, IQ2_XS is 11.3 GB, IQ2_S is 12.0 GB, IQ2_M is 13.2 GB, Q2_K is 13.8 GB and 20 more.
What context length does MiniCPM V 4.7 35B A3B support?
The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 0.7 GB at 8K and 2.7 GB at 32K.
Where can I download MiniCPM V 4.7 35B A3B GGUF?
The files are on Hugging Face in the bartowski/MiniCPM-V-4.7-35B-A3B-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
What license does MiniCPM V 4.7 35B A3B use?
Hugging Face lists the license as apache-2.0. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.
Can I run MiniCPM‑V 4.7 35B‑A3B on 8 GB VRAM?
Use the Q4_K_M quant; it estimates 23.5 GB memory at 8 k context, which is above 8 GB but can be run with off‑load or lower context.
Does the model support video input?
The model supports image input; video support is not listed in the provided facts.
What license applies to this model?
The model is released under the Apache‑2.0 license.