Vision

Llama 3.2 Community LicenseOpen weights where releasedUpdated August 2026

Llama 3.2 Vision

Llama 3.2 Vision models are positioned for image-aware assistant and multimodal application workflows.

Meta · Llama

Editorial review

Reviewed byOpenSourcesAI EditorialLast updatedAugust 2026SourcesCanonical artifact, Representative model card

Model checkpoints, context windows, provider support, local runtime compatibility, and license terms can change quickly. Verify the exact model card before production or commercial use.

Best for

Vision-language experiments, screenshot reasoning, and multimodal app prototypes.

Who should use it

  • Vision-language experiments, screenshot reasoning, and multimodal app prototypes.
  • Builders who want local or self-hosted testing options.

Common workflows

  • Vision-language assistant workflows
  • vision workflows
  • multimodal workflows
  • assistant workflows

Deployment and hardware notes

The representative 11B Vision Instruct checkpoint is 10.7B parameters, so plan for roughly 6.5 GB at Q4_K_M; the 90B sibling is multi-GPU territory.

License and usage notes

Llama 3.2 Community License. Open weights where released. Verify the exact model card and license terms for the checkpoint or hosted provider you use.

Strengths

  • Open weights where released model option for Llama workflows.
  • Vision-language experiments, screenshot reasoning, and multimodal app prototypes.
  • Access must be requested on the model page before the weights download — a gate that stops otherwise-working pipelines at first run.

Limitations

  • Fixed 560-pixel tiling with a four-tile ceiling caps effective resolution, so dense screenshots and small print fare worse than on the dynamic-resolution Qwen models. The weights are gated on Hugging Face and carry the Llama 3.2 Community Licence rather than an open-source one, and the licence excludes multimodal use in the EU.
  • The representative 11B Vision Instruct checkpoint is 10.7B parameters, so plan for roughly 6.5 GB at Q4_K_M; the 90B sibling is multi-GPU territory.
  • Context window and limits: 131,072 tokens (RoPE-extended).
  • Verify the exact model card, provider docs, license, and serving support before production use.

Local workflow notes

Access must be requested on the model page before the weights download — a gate that stops otherwise-working pipelines at first run.

Local runtimes: Ollama where supported, LM Studio where supported, llama.cpp where supported, Transformers

Platforms: Windows, macOS, Linux

Vision spec

This page covers a family of checkpoints. The figures below describe Llama-3.2-11B-Vision-Instruct, its representative release — other sizes in the family differ.

MemoryVaries by size; released as 11B and 90B vision checkpoints — check the exact one you plan to runImage inputUp to 4 tiles of 560x560 pixelsContext131,072 tokens (RoPE-extended)

Meta's multimodal Llama: a vision adapter bolted onto a Llama 3.1 text tower, which is why the text behaviour is familiar while the image path is comparatively rigid.

Sources to verify

Canonical artifactRepresentative model card

Related resources

Continue with model source notes, local tools, and implementation guides related to this model.

Llama family hubAI tools directoryModel selection guideModel source notesModels directory
HardwareVaries by size; released as 11B and 90B vision checkpoints — check the exact one you plan to runRuntimeOllama or LM Studio where supported, llama.cpp, Transformers, vLLMContext131,072 tokens (RoPE-extended)Last updated2026
Canonical artifact →

Model ecosystem connections

Use these next-step links to move from this profile into related tools, comparisons, guides, stacks, and curated shortlists.

Recommended runtimes and tools

Ollama →LM Studio →llama.cpp →Transformers →vLLM →

Setup and deployment

Check if this model fits your GPU →What is quantization? GGUF, Q4, Q8 explained →What is VRAM and how much do you need? →

Related model pages

Llama family hub →Llama 3.3 70B Instruct →Llama 3 70B →Llama 3 8B Instruct →PaliGemma 2 →Phi-3.5 Vision →Llama 3.3 70B →

Guides, stacks, and comparisons

AI tools directory →Model selection guide →All model comparisons →

Related models

Llama 3.3 70B InstructGeneral assistant workflows, app prototypes, and Llama-family baseline comparisons.Llama 3 70BBuilders who want a widely supported open-weight chat model with broad runtime compatibility.Llama 3 8B InstructLocal baseline comparisons and lightweight app prototypes.PaliGemma 2Vision-language app prototypes and multimodal evaluation.Phi-3.5 VisionSmall vision-language experimentsLlama 3.3 70B70B open-weight model in the Llama family.

Ready to run this model locally?

Find a compatible interface in our Local AI Tools directory →