Llama 3.2 Vision 11B is Meta's accessible open multimodal model — handles image-grounded QA, chart reading, and document understanding well. Fits in 16 GB VRAM at Q4_K_M. Competitive with GPT-4V on common tasks but lags on dense-text OCR and very detailed visual reasoning where the 90B variant pulls ahead.
Llama 3.2 Vision 11B needs at minimum 6.7 GB of VRAM at Q4_K_M quantization (24.2 GB at FP16). The cheapest GPU that comfortably fits with KV-cache headroom is the Intel Arc B570 (10 GB VRAM, $219 MSRP). Community benchmark submissions are open. This model fits a single consumer GPU under 48 GB, so a one-card build works.
Weights only. Add ~20-30% for KV cache at typical context lengths.
| Quant | VRAM | Quality | Notes |
|---|---|---|---|
| FP16 | 24.2 GB | Reference | Training-precision weights |
| Q8_0 | 12.1 GB | Near-lossless | 8-bit, ~0.1% perplexity hit |
| Q6_K | 9.8 GB | Very high | 6-bit, near Q8 quality |
| Q5_K_M | 8.3 GB | High | Strong middle ground |
| Q4_K_M | 6.7 GB | Balanced (recommended) | Default for local deployments |
| Q4_0 | 6.8 GB | Legacy 4-bit | Older GGUF, kept for compatibility |
| Q3_K_M | 5.2 GB | Lossy | When VRAM is very tight |
| Q2_K | 3.9 GB | Extreme | Rescue option, quality degrades visibly |
Filtered from MyAIHardware's GPU database. FP16 needs 24.2 GB; Q4_K_M needs 6.7 GB.
We don't have community benchmarks for this exact model yet.
Official weights + popular runtime tags.
ollama pull llama3.2-vision:11blms get bartowski/Llama-3.2-11B-Vision-Instruct-GGUFStep-by-step guides that use this model.
Use these in papers, blog posts, or internal docs.
@misc{llama-3-2-vision-11b-2024,
title={Llama 3.2 Vision 11B},
author={Meta AI},
year={2024},
url={https://huggingface.co/meta-llama/Llama-3.2-11B-Vision-Instruct}
}Meta AI (2024). Llama 3.2 Vision 11B [Model card]. Hugging Face. https://huggingface.co/meta-llama/Llama-3.2-11B-Vision-Instruct
Llama 3.2 Vision 11B (Llama 3.2, Meta AI, 2024). Available at https://huggingface.co/meta-llama/Llama-3.2-11B-Vision-Instruct.