What are the hardware requirements for Llama 3.2 90B Vision Instruct?

Llama 3.2 90B Vision Instruct requires approximately 178.2 GB of VRAM. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Llama 3.2 90B Vision Instruct and how large is it?

Llama 3.2 90B Vision Instruct is developed by unsloth, a model with 90b 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 Llama 3.2 90B Vision Instruct?

Llama 3.2 90B Vision Instruct is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Llama 3.2 90B Vision Instruct by unsloth — VRAM 178.2GB

Name: Llama 3.2 90B Vision Instruct
Author: unsloth

Llama 3.2 90B Vision Instruct is an open-source language model by unsloth. Features: 90b LLM, VRAM: 178.2GB, License: llama3.2, Instruction-Based, LLM Explorer Score: 0.15.

Base model:finetune:meta-llama... Base model:meta-llama/llama-3.... Conversational Deploy:azure En Endpoints compatible Facebook Image-text-to-text Instruct Llama Llama-3 Meta Mllama Multimodal Region:us Safetensors Sharded Tensorflow Unsloth Vision

Model Card on HF 🤗: https://huggingface.co/unsloth/Llama-3.2-90B-Vision-Instruct

Llama 3.2 90B Vision Instruct Parameters and Internals

LLM Name	Llama 3.2 90B Vision Instruct
Repository 🤗	https://huggingface.co/unsloth/Llama-3.2-90B-Vision-Instruct
Base Model(s)	meta-llama/Llama-3.2-90B-Vision-Instruct meta-llama/Llama-3.2-90B-Vision-Instruct
Model Size	90b
Required VRAM	178.2 GB
Updated	2026-08-10
Maintainer	unsloth
Model Type	mllama
Instruction-Based	Yes
Model Files	4.6 GB: 1-of-37 4.7 GB: 2-of-37 5.0 GB: 3-of-37 5.0 GB: 4-of-37 4.7 GB: 5-of-37 4.7 GB: 6-of-37 4.7 GB: 7-of-37 5.0 GB: 8-of-37 5.0 GB: 9-of-37 4.7 GB: 10-of-37 4.7 GB: 11-of-37 4.7 GB: 12-of-37 5.0 GB: 13-of-37 5.0 GB: 14-of-37 4.7 GB: 15-of-37 4.7 GB: 16-of-37 4.7 GB: 17-of-37 5.0 GB: 18-of-37 5.0 GB: 19-of-37 4.7 GB: 20-of-37 4.7 GB: 21-of-37 4.7 GB: 22-of-37 5.0 GB: 23-of-37 5.0 GB: 24-of-37 4.7 GB: 25-of-37 4.7 GB: 26-of-37 4.7 GB: 27-of-37 5.0 GB: 28-of-37 5.0 GB: 29-of-37 4.7 GB: 30-of-37 4.7 GB: 31-of-37 4.7 GB: 32-of-37 5.0 GB: 33-of-37 5.0 GB: 34-of-37 4.7 GB: 35-of-37 4.7 GB: 36-of-37 4.9 GB: 37-of-37
Supported Languages	en
Model Architecture	MllamaForConditionalGeneration
License	llama3.2
Model Max Length	131072
Transformers Version	4.45.0
Tokenizer Class	PreTrainedTokenizerFast
Padding Token	<\|finetune_right_pad_id\|>
Torch Data Type	bfloat16

Rank the Llama 3.2 90B Vision Instruct 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 Llama 3.2 90B Vision Instruct

Best Alternatives	Context / RAM	Downloads	Likes
Llama 3.2 90B Vision Instruct	0K / 178.2 GB	30	6
...a 3.2 90B Vision Instruct 4bit	0K / 49.9 GB	318	4
...sion Instruct Unsloth Bnb 4bit	0K / 50.6 GB	17	2
...2 90B Vision Instruct Bnb 4bit	0K / 49.1 GB	361	19

Note: green Score (e.g. "73.2") means that the model is better than unsloth/Llama-3.2-90B-Vision-Instruct.