MI

Pixtral

MistralGGUF by EnlistedGhostUpdated 2026-08-25
Download on Hugging Face
Parameters12.2B
Context1000K
Layers40
Architecturellama
Smallest quant5.1 GB
Quick answer

Pixtral needs about 9.5 GB of memory at Q4_K_M (a 7.7 GB file) with an 8K context window, so it fits on a 12 GB GPU. This repository has 21 quants, from 5.1 GB (Q2_K) to 24.5 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 5.4 GB.

Pixtral-12B-2409: 12B multimodal model with 400M vision encoder

Pixtral-12B-2409 is a multimodal model with 12B parameters and a 400M parameter vision encoder. It supports variable image sizes and a sequence length of 128k. The model is licensed under Apache 2.0.

Good at

  • Multimodal inference
  • Variable image size handling
  • Long‑context text generation

Limits

  • No moderation mechanisms
  • Requires 8k+ context for accurate memory estimates
  • No audio or video support

Best for

  • Vision‑enabled chatbots
  • Image‑captioning pipelines
  • Multimodal research prototyping

Who it’s for. Developers and researchers building multimodal applications who need a large‑scale vision‑language model. The model is suitable for production inference when combined with vLLM or mistral_inference.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q2_K5.1 GB 7.0 GB
Fits Get file →
IQ3_M5.7 GB 7.6 GB
Tight Get file →
Q3_K_S5.8 GB 7.6 GB
Tight Get file →
Q3_K_M6.3 GB 8.2 GB
Too big Get file →
Q2_K_M6.4 GB 8.3 GB
Too big Get file →
Q3_K_L6.8 GB 8.7 GB
Too big Get file →
Q2_K_L7.0 GB 8.8 GB
Too big Get file →
IQ4_XS7.0 GB 8.8 GB
Too big Get file →
Q3_K_XL7.1 GB 9.0 GB
Too big Get file →
Q4_K_S7.3 GB 9.1 GB
Too big Get file →
Q4_K_M7.7 GB 9.5 GB
Too big Get file →
Q2_K_XL7.9 GB 9.8 GB
Too big Get file →
Q5_K_S8.6 GB 10.5 GB
Too big Get file →
Q5_K_M8.8 GB 10.7 GB
Too big Get file →
Q4_K_XL9.4 GB 11.2 GB
Too big Get file →
Q6_K10.1 GB 11.9 GB
Too big Get file →
Q6_K_L10.4 GB 12.2 GB
Too big Get file →
Q5_K_XL10.4 GB 12.3 GB
Too big Get file →
Q6_K_XL11.3 GB 13.2 GB
Too big Get file →
Q8_013.0 GB 14.9 GB
Too big Get file →
BF1624.5 GB 26.3 GB
Too big Get file →

For 8‑10 GB VRAM, use Q4_K_M (file size 7.7 GB, estimated 9.5 GB memory at 8k context). For 12‑14 GB VRAM, Q6_K (10.06 GB, 11.9 GB memory) or Q6_K_XL (11.35 GB, 13.2 GB memory) are suitable. Larger budgets can use Q8_0 or BF16.

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_S (5.8 GB file, about 7.6 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).
  • 24 GB VRAM: the best quant that fits is Q8_0 (13.0 GB file, about 14.9 GB in use).
  • 32 GB VRAM: the best quant that fits is BF16 (24.5 GB file, about 26.3 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 7.7 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 5.1 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

Architecturellama
Parameters12.2B
Layers40
Trained context1,024,000 tokens
KV cache (fp16)160 KB per token

Source

Quantized byEnlistedGhost
FamilyMistral
Licenseapache-2.0
Downloads (HF)2,025
Last updated2026-08-25

Frequently asked questions about Pixtral

How much VRAM does Pixtral need?

About 9.5 GB at Q4_K_M with an 8K context window. That is the 7.7 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_S: the 5.8 GB file needs about 7.6 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 5.1 GB, IQ3_M is 5.7 GB, Q3_K_S is 5.8 GB, Q3_K_M is 6.3 GB, Q2_K_M is 6.4 GB, Q3_K_L is 6.8 GB and 15 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 EnlistedGhost/Pixtral-12B-2409-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?

The smallest quant that fits 8 GB VRAM is Q4_K_M, which requires about 9.5 GB memory at 8k context, so 8 GB VRAM is insufficient for full inference.

How do I use Pixtral with Ollama?

Download the GGUF file from the repository and load it directly in Ollama; both vision and text work natively without extra configuration.

Does Pixtral support variable image sizes?

Yes, the model was trained with interleaved image and text data and supports variable image sizes during inference.

Similar models

Browse all →

More from EnlistedGhost

All by EnlistedGhost →

More Mistral models

All Mistral →

Other medium models

All medium models →