Muse Glimmer 30B is the open-weight dense multimodal model from Meta Superintelligence Labs for local agentic tasks, distilled from Muse Spark. About 29.6B parameters including a 1.8B perception encoder · 52 layers · text and image in, text out · 131,072 token context · reasoning strength low to xhigh · knowledge cutoff 2026-01-04 · trained on data from more than 100 languages. Weights on Hugging Face since 2026-08-09 under Apache 2.0 (BF16, two 4-bit builds for 24 GB and 32 GB VRAM, DFlash drafter). Research paper published 2026-10-19.
No benchmark data available yet.
- Typemultimodal
- Context131K tokens (~66 books)
- ReleasedAug 2026
- LicenseOpen Source
- StatusActive
- Cost / Message~$0.002
Frequently Asked Questions
Related Models
Qwen3 Coder Next FP8 · AlibabaQwen2.5 Coder 32B Instruct AWQ · AlibabaSpeaker Diarization Community 1 · PyannoteJais 3 0.9B · MBZUAIGemma 3 12B (free) · Google DeepMindBenchmarks
Key facts · as of 2026-10-20
- Muse Glimmer 30B by Meta. Not enough public benchmark scores to rank yet.
- List price $0.30 input · $1.20 output per 1M tokens (as of 2026-10-20).
How to cite · data as of 2026-10-20
Muse Glimmer 30B · benchmarks, pricing and providers. BenchGecko, data as of 2026-10-20. https://benchgecko.ai/model/muse-glimmer-30b
Credit "Source: BenchGecko" with a link. Prices per provider and Gecko Tests are BenchGecko data (CC BY 4.0); benchmark scores keep their original source, listed in the JSON. JSON · llms.txt · MCP