Meta's open-weights 30B dense model for local agents, tuned for tool use and long tasks, Apache 2.0 licensed.
Updated EU status checked Last shown
Sources: the vendor's own pages (linked above) and creator videos at the linked timestamps. EU status is information, not legal advice; see how we check.
Muse Glimmer is an open 30B-parameter dense language model from Meta, positioned for always-on local agents. Meta's page says it runs on a single GPU and is tuned for tool use, long-running tasks and failure recovery. Creators report weights of roughly 55–60 GB at full precision, with a quantized version also available.
Meta's own benchmarks show it ahead of similar-size open models such as Gemma 4 31B and Qwen 3.6 27B on many tests. Independent views are mixed: one reviewer, citing Artificial Analysis, says it does not beat Qwen 3.6, and another reports a score of 51 on Terminal Bench, well below frontier models. Its BusyBench output was described as poor.
It is released under Apache 2.0 and is intended for developers who want to self-host an agentic model on a desktop-class GPU. Meta's page lists run options, build resources and cookbooks, and a Model API exists on dev.meta.ai, though details of its pricing and country availability were not in the provided material.
Watch the moment: the exact second top creators show it. Videos load only when you press play, via youtube-nocookie.com.
Open 30B dense model under Apache 2, around 60 GB. Meta's benchmarks look cherry-picked; Artificial Analysis shows it does not beat Qwen 3.6, so the creator calls it underwhelming.
Meta's open-weights 30B-parameter model meant for on-device use, fitting a desktop GPU. It scores 51 on Terminal Bench, well below frontier models, and the creator is glad Meta is releasing open models again.
Meta's small open-weight agentic model for local use. It beats similar-size models such as Gemma 4 31B and Qwen 3.6 27B on many benchmarks. Full precision needs about 55 GB, and a quantized version is available. Its BusyBench output was poor.
Closest neighbours by meaning, not just by category.
Model · Meta
Meta's open-weight 816B-parameter mixture-of-experts model (42B active, 1M context) in the Muse model family.
Model · Meta
Closed-source Meta model for coding and agentic workflows with 1M-token context and multimodal input, offered via API.
Model · Meta
Meta's Muse Spark 1.3 language model, described by creators as near-frontier on benchmarks.
Model
35B Mixture-of-Experts open-weights model from InternScience for long-horizon agentic search, engineering and tool use.
Model · Meta
Meta's frontier model for personal and professional AI agents, with Active Memory and Active Focus; successor to Muse Spark.
Model
Open-weights 35B-A3B agentic model post-trained from Qwen3.6-35B-A3B for search, coding and scientific research.
