Meta Superintelligence Lab

Muse-Glimmer-30B · live demo

A 30B multimodal (image-text-to-text) model. See the model card at meta-models/Muse-Glimmer-30B. This Space is a static page: the demo below runs on the model author's Gradio Space, and you can also chat yourself with your own Hugging Face token.

image-text-to-text apache-2.0 30B params transformers ZeroGPU

1 · Try it now (no setup)

Embedded, fully working chat running on the community ZeroGPU Space. If it doesn't load, open akhaliq-muse-glimmer-30b.hf.space directly.

2 · Or run it with your own token

Calls the Hugging Face Inference Providers router from your browser. Your token is kept in localStorage only — it never leaves your device except to Hugging Face. A token with Inference Providers credits (or a PRO plan) is required. Get one at settings/tokens.

Static Space · Built from meta-models/Muse-Glimmer-30B. For a native ZeroGPU Gradio version of this Space, see app.py in this repo (requires PRO or a community grant).