What are the hardware requirements for Pixtral 12B 4bit?

Pixtral 12B 4bit requires approximately 7.2 GB of VRAM. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Pixtral 12B 4bit and how large is it?

Pixtral 12B 4bit is developed by mlx-community, a model with 12b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

Where can I download or evaluate Pixtral 12B 4bit?

Pixtral 12B 4bit is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Pixtral 12B 4bit by mlx-community — VRAM 7.2GB

Name: Pixtral 12B 4bit
Author: mlx-community

Pixtral 12B 4bit is an open-source language model by mlx-community. Features: 12b LLM, VRAM: 7.2GB, Quantized.

4bit Conversational Endpoints compatible Image-text-to-text Mlx Pixtral Quantized Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/mlx-community/pixtral-12b-4bit

Pixtral 12B 4bit Parameters and Internals

LLM Name	Pixtral 12B 4bit
Repository 🤗	https://huggingface.co/mlx-community/pixtral-12b-4bit
Model Size	12b
Required VRAM	7.2 GB
Updated	2026-08-08
Maintainer	mlx-community
Model Type	pixtral
Model Files	5.4 GB: 1-of-2 1.8 GB: 2-of-2
Quantization Type	4bit
Model Architecture	LlavaForConditionalGeneration
Tokenizer Class	PreTrainedTokenizerFast
Torch Data Type	bfloat16

Rank the Pixtral 12B 4bit capabilities

Have you tried this model? Rate its performance — this feedback helps the ML community find the right model for their needs.

Instruction Following and Task Automation
Factuality and Completeness of Knowledge
Censorship and Alignment
Data Analysis and Insight Generation
Text Generation
Text Summarization and Feature Extraction
Code Generation
Multi-Language Support and Translation

Best Alternatives to Pixtral 12B 4bit

Best Alternatives	Context / RAM	Downloads	Likes
...tral 12B 2409 Unsloth Bnb 4bit	0K / 9.2 GB	7204	14
Pixtral 12B 8bit	0K / 13.5 GB	213	8
Pixtral 12B Bf16	0K / 25.3 GB	74	1

Note: green Score (e.g. "73.2") means that the model is better than mlx-community/pixtral-12b-4bit.