Meta releases Muse Glimmer, a 30B open-weight model for local AI agents, under Apache 2.0
Announced · Published · By Contractor AI Shop
Meta released Muse Glimmer, a 30-billion-parameter open-weight model it says is built for local agent work and small enough to run on a Mac or PC with one consumer GPU. The weights are under the Apache 2.0 license. Ollama added it the same day.
Key facts
- Meta announced Muse Glimmer on August 10, 2026 and released the weights under the Apache 2.0 license. Meta
- Meta describes it as a 30-billion-parameter model that is small enough to run on a Mac or a PC with a single consumer GPU. Meta
- TechCrunch reported it is designed to power AI agents locally on consumer hardware. TechCrunch
- Ollama lists a 128K+ context length and says you can run it with 'ollama run muse-glimmer'. Ollama
Our take: what it means for contractors
This is one of the more permissive licenses for a capable model, so you can use it in your business without per-seat fees. Try it in Ollama on a machine you already own. Give it a real task, like summarizing a spec section or drafting an RFI, and compare the result with what you get from a paid chatbot.
What we don't know yet
- Real-world speed on typical contractor hardware
- How it compares to paid models on construction documents
- Exact memory needs for each quantized version
Sources
- Meta · Introducing Muse Glimmer: An Open Agentic Model That Runs on Your Device · 2026-08-10 · Primary
- TechCrunch · Meta's new Glimmer AI model offers a hint at Zuckerberg's personal intelligence vision · 2026-08-10 · Secondary
- CNBC · Meta launches Muse Glimmer open-weight AI model · 2026-08-10 · Secondary
- Ollama · Muse Glimmer from Meta Superintelligence Labs is now available · Secondary