Tag

#vision-language-model

1 story tagged vision-language-model.

Muse Glimmer open-source multimodal AI model parameter breakdown: 2 billion vision encoder and 28 billion text decoder, totaling 30 billion parameters.

Meta's Muse Glimmer ships as open-source agentic VLM

Muse Glimmer is Meta's 30B open-source multimodal model with hybrid attention and self-deploying agents. Day-0 support ships across major inference stacks.

Lars Cornelissen · Aug 10, 2026