Meta open-weights 816B Muse Spark Open, Apple tunes it for MLX on Mac
rohanpaul_ai · x · 2026-10-16
Meta has released open weights for Muse Spark Open, an 816B-parameter model that Apple tuned for its MLX framework and that runs on two linked Mac Studios. The MoE model routes each token to 8 of 256 experts (42B active), so each step reads only a small slice of memory. The hard part is holding the model: the 4-bit NVFP4 build needs at least 700GB of GPU memory. Each M5 Ultra Mac Studio offers up to 512GB unified memory, so two machines hold the model with 1.2TB/s of bandwidth feeding each token. It tops the MCP-Atlas benchmark.
More from Infra
- Google's Willow chip cuts quantum errors 40x, raising Q-Day concerns for Bitcoin — jamestagg · 2026-10-16
- Vibe Coding a Data Pipeline Is Easy; Running It at Billions of Rows Is Not — Zachly · 2026-10-16
- Large data centers employ only ~50-150 permanent on-site staff, Microsoft says — brezina · 2026-10-16
- Prediction: Opus 5.5-level models on laptops by 2027, backed by Qwen 27B data — cephaloform · 2026-10-16
- Oracle down 12% in a week, nearly 50% off June high as OpenAI exposure spooks market — alvelda · 2026-10-16
- US data center moratoriums top 400 — an AI bull explains why they're actually good for AI — kevinsxu · 2026-10-16