facebookresearch/chameleon

Health v2 · maintenance only

Flags: no_releases archived no_license

Adoption not part of the score

2103 stars · 118 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

Repository for Meta Chameleon, an early-fusion token-based mixed-modal foundation model that understands and generates interleaved images and text. It provides GPU inference code, a browser-based multimodal input/output viewer, a lightweight miniviewer, and evaluation prompts for the released 7B and 30B checkpoints.

Use cases

• run inference with a mixed-modal image-and-text foundation model
• generate images and text interleaved from a single model
• visualize multimodal model inputs and outputs in a browser
• evaluate a multimodal model on visual question answering and captioning
• experiment with early-fusion token-based multimodal architectures
• download and serve Meta Chameleon checkpoints locally

When to choose

• you need a single model that both understands and generates images and text
• you want to reproduce or study the Chameleon paper's early-fusion approach
• you have CUDA GPUs and want fast local inference of the 7B or 30B checkpoints
• you need a tool to inspect mixed-modal prompts and generations

When to avoid

• you only need text-only LLM inference with broad ecosystem support
• you lack a CUDA-capable GPU and cannot use the HuggingFace implementations
• you need a production-ready, actively maintained multimodal API
• you cannot accept the custom research license or checkpoint access requirements

Facets

library · maturity maintenance

llm-inference machine-learning image-processing nlp large-language-models artificial-intelligence deep-learning machine-learning python multimodal foundation-model early-fusion image-generation text-generation meta-ai research-model inference gpu docker linux

Member repositories

For agents

markdown · JSON · MCP: product_card(name="facebookresearch/chameleon")