Meta Releases Muse Glimmer: A 30B Open-Weight Agentic Model Running on 24GB VRAM
alexandr_wang · x · 2026-08-10
Meta Superintelligence Labs has introduced Muse Glimmer, a 30-billion-parameter open-weight agentic model released under the Apache 2.0 license.
- Local Execution: Optimized for always-on local agent workflows, it runs on a single consumer GPU with 24GB VRAM. It supports local coding, function calling, and LLM-as-a-judge evaluation.
- Inference Optimization: Built with a custom architecture, it quantizes weights to 4-bit to fit under 20GB. It uses a dflash drafter to propose token blocks verified in parallel by the main model, ensuring responsiveness.
- Agentic Capabilities: Functions as a fully capable agent with planning, tool calling, result checking, and failure recovery, rivaling much larger models.
- Ecosystem: Weights are available on Hugging Face, with integrations for Ollama, vLLM, llama.cpp, and others rolling out this week.
An open-weight version of another model, Muse Spark 1.2, is also coming soon.
Related event: Meta Open-Sources 30B Muse Glimmer for On-Device Agents(108 posts)→
More from Infra
- Amazon Proposes Per-Request Validated Evidence Caching for Production RAG — _reachsumit · 2026-10-16
- LinkedIn's TransX report: sequence transduction for recsys, +6.02% CTR in A/B tests — _reachsumit · 2026-10-16
- llama.cpp PR boosts CPU inference speed with merged AVX2 tiled accumulators — jacek2023 · 2026-10-16
- Zero's next-gen IVM engine runs in a single DO, live-querying a 100GiB R2 database — threepointone · 2026-10-16
- Meme: driving to the chip factory in 2028 captures AI's compute anxiety — beffjezos · 2026-10-16
- One volunteer maintains the time zone file behind 4 billion phones — danbri · 2026-10-16