Muse Spark Open
Points and comments are a snapshot, not live.
Meta releases Muse Spark Open, an 816B-parameter open-weight model optimized for Apple Silicon.
Muse Spark Open is Meta's new open-weight model with 816B total parameters (42B active) and a 1M-token context window, designed for coding and agentic work. It runs on two Apple M5 Ultra Mac Studios using MLX, generating over 36 tokens per second on a four-Mac cluster. Weights are available on Hugging Face and other platforms, with support for multiple inference engines. The model excels in benchmarks for agentic, coding, and reasoning tasks. It is released under the Muse Model License with limited commercial terms and includes safety post-training.
What commenters are saying
Commenters praised Meta's release of an open-weight model and its broad inference engine support. Several noted the high hardware cost, with a single M5 Ultra Mac Studio at $11k. A camp argued Meta will dominate consumer AI via ecosystem lock-in, drawing skepticism about stickiness given Muse's novelty and branding as OpenClaw reskin. Others noted NPUs are unsuitable for large models like this. The thread highlighted Meta's strategic effort to standardize its model and infrastructure.