Pixtral needs about 9.3 GB of memory at Q4_K_M (a 7.5 GB file) with an 8K context window, so it fits on a 12 GB GPU. This repository has 11 quants, from 4.8 GB (Q2_K) to 13.0 GB (Q8_0). Longer context needs more: at 32K tokens the KV cache alone adds about 5.4 GB.
Pixtral 12B GGUF – 12.2B‑parameter vision‑enabled model
Pixtral is a 12.2B‑parameter model from the Mistral family, built on the llama architecture. It supports vision tasks and is available in multiple GGUF quantizations. The model is licensed under apache-2.0.
Good at
- Vision and image‑text tasks
- Large context handling
- High‑parameter density
Limits
- No vision support in non‑vision quantizations
- Requires 8k context for memory estimates
- License is apache-2.0
Best for
- Image captioning
- Multimodal chat
- Visual question answering
Who it’s for. Developers building multimodal applications who need a high‑parameter model with vision capabilities. The model is suitable for research and production use where the apache-2.0 license is acceptable.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q2_K | 4.8 GB | 6.6 GB | Fits | Get file → | |
| Q3_K_S | 5.5 GB | 7.4 GB | Tight | Get file → | |
| Q3_K_M | 6.1 GB | 7.9 GB | Tight | Get file → | |
| Q3_K_L | 6.6 GB | 8.4 GB | Too big | Get file → | |
| IQ4_XS | 6.8 GB | 8.6 GB | Too big | Get file → | |
| Q4_K_S | 7.1 GB | 9.0 GB | Too big | Get file → | |
| Q4_K_M | 7.5 GB | 9.3 GB | Too big | Get file → | |
| Q5_K_S | 8.5 GB | 10.4 GB | Too big | Get file → | |
| Q5_K_M | 8.7 GB | 10.6 GB | Too big | Get file → | |
| Q6_K | 10.1 GB | 11.9 GB | Too big | Get file → | |
| Q8_0 | 13.0 GB | 14.9 GB | Too big | Get file → |
For 8GB VRAM, use Q4_K_M (file size 7.48 GB, estimated memory 9.3 GB). For 12GB VRAM, Q5_K_M (8.73 GB, 10.6 GB) or Q6_K (10.06 GB, 11.9 GB) are suitable. For 16GB VRAM, Q8_0 (13.02 GB, 14.9 GB) provides the highest quality.
What GPU do you need for Pixtral?
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: no quant fits at an 8K context.
- 6 GB VRAM: no quant fits at an 8K context.
- 8 GB VRAM: the best quant that fits is Q3_K_M (6.1 GB file, about 7.9 GB in use).
- 12 GB VRAM: the best quant that fits is Q6_K (10.1 GB file, about 11.9 GB in use).
- 16 GB VRAM: the best quant that fits is Q8_0 (13.0 GB file, about 14.9 GB in use).
Which Pixtral quant should you choose?
Q4_K_M is the usual starting point for Pixtral: it balances file size and quality well. Here it is a 7.5 GB file.
The Q8_0 file is 13.0 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 Q2_K at 4.8 GB, with the quality trade-offs that come with fewer bits. Use it only when memory is very tight. Expect a clear drop in quality, which is usually worse on small models and easier to live with on large ones.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Pixtral
How much VRAM does Pixtral need?
About 9.3 GB at Q4_K_M with an 8K context window. That is the 7.5 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 Pixtral on an 8 GB GPU?
Yes, with Q3_K_M: the 6.1 GB file needs about 7.9 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of Pixtral GGUF?
It depends on the quant: Q2_K is 4.8 GB, Q3_K_S is 5.5 GB, Q3_K_M is 6.1 GB, Q3_K_L is 6.6 GB, IQ4_XS is 6.8 GB, Q4_K_S is 7.1 GB and 5 more.
What context length does Pixtral support?
The GGUF header lists a trained context length of 1,024,000 tokens (1000K). The KV cache grows with the context you use: about 1.3 GB at 8K and 5.4 GB at 32K.
Where can I download Pixtral GGUF?
The files are on Hugging Face in the mradermacher/pixtral-12b-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 Pixtral 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 Pixtral on 8 GB VRAM?
Yes, the Q4_K_M quant uses about 9.3 GB of VRAM at 8k context, which fits within 8 GB with some headroom.
Does Pixtral support vision tasks?
Yes, Pixtral is designed for vision and image‑text tasks, as indicated by its detected abilities.