← October 16, 2026 briefing

Meta's Muse Spark Open runs locally on Macs with Apple's help, with day-one support on AWS SageMaker JumpStart, Transformers and NVIDIA GB300

Announced October 15, 2026

Verified (fact)

Deployment options have expanded for Muse Spark Open, the 816B MoE (42B active) open-weight model released by Meta. Meta and Apple worked together to run the model locally on Apple Silicon. Meta said it can run on two M5 Ultra Mac Studios using the latest MLX framework. AWS made the model deployable to endpoints in customers' own accounts via SageMaker JumpStart, and Hugging Face Transformers v5.20.0 supports BF16 and NVFP4 checkpoints. NVIDIA published guidance on scaling agentic coding workflows on GB300 NVL72.

Needs fact-check (claim)

Meta claims it is "the best American open-weight model to date" and "the first model in its class that can run on a desktop."

Why it matters (interpretation)

A frontier-class open-weight model is supported from day one across cloud, data center GPUs, and local Macs. This sharply lowers the barrier for companies to start using it in their own environments.

Sources

Story thread

  1. 2026-10-15 Meta releases 'Muse Spark Open', an 816B MoE open-weight model with 42B active and 1M-token context for coding and agents
  2. 2026-10-16 Meta's Muse Spark Open runs locally on Macs with Apple's help, with day-one support on AWS SageMaker JumpStart, Transformers and NVIDIA GB300

More about Meta