Meta Releases 30B Open-Weight Multimodal Model Muse Glimmer, Runs on 24GB VRAM
spillai · x · 2026-08-12
Meta's Superintelligence Labs has released Muse Glimmer, an open-weight multimodal model. Licensed under Apache 2.0, the 30B parameter model is designed for consumer hardware and can run entirely locally on 24GB VRAM GPUs or Macs.
For developers building on open-weight Vision-Language Models (VLMs), the 30B parameter count offers a sweet spot. It is large enough to handle complex multimodal reasoning while remaining compact enough for teams to deploy without renting expensive compute clusters. This makes it highly viable for always-on agentic workflows that require continuous model execution.
Related event: Meta Releases 30B Open-Source Muse Glimmer for On-Device Agents(108 posts)→
More from Models
- User hopes for Claude V4 Pro this week, notes delay from mid-July to July 31 — teortaxesTex · 2026-08-12
- Gemini Confuses Its Own Creator Google with OpenAI — Ancient-Tomato-5226 · 2026-08-12
- Train Real Language Models from Scratch Directly in Your Browser — chrisgrayson · 2026-08-12
- LlamaIndex Launches ExtractBench: 4,869 Pages of Complex Docs Across 8 Domains — llama_index · 2026-08-12
- Claude Opus 5 Drifts Into British Spellings, Cites Context as Precedent — ericm272 · 2026-08-12
- Safety Guardrails Hinder Bug Fixing Due to Keyword Triggers — xeophon · 2026-08-12