Muse Glimmer 30B
Muse Glimmer 30B is an open-weight, 30B-parameter multimodal model from Meta Superintelligence Labs, released under Apache 2.0 and optimized for always-on local agent workflows on a single consumer GPU or Apple Silicon. It features a 131K-token context window, training in 100+ languages, DFlash acceleration, and is distilled from Muse Spark for coding, evaluation, and agentic tasks, with integrations for llama.cpp, MLX, and ExecuTorch on Hugging Face.