Apple ML Research: Muse Spark Open with MLX on Mac Studio
What happened
Apple’s MLX framework now supports running the large language model Muse Spark Open, which has nearly a trillion parameters, on Mac Studio machines. Using the latest MLX release, a single Mac Studio with up to 512 GB of unified memory can run the model, and a cluster of four Mac Studios can generate over 36 tokens per second. The framework leverages Apple silicon’s unified memory and GPU Neural Accelerators, and can scale across multiple Macs via RDMA over Thunderbolt. Performance benchmarks show that the newer M5 Ultra chip offers up to 4.6× faster time‑to‑first‑token compared to the M3 Ultra, while generation speed is limited by memory bandwidth.
Why it matters
This makes it possible for developers to experiment with cutting‑edge, trillion‑parameter models on local hardware without needing cloud resources. It also demonstrates that Apple’s silicon can handle very large contexts, which could benefit applications that require extensive text processing.
Reality check
An automated, sceptical read of the sourceIncremental. The update builds on existing MLX capabilities and adds support for a single new large model, without introducing fundamentally new technology.
Is this a big deal?
Get stories like this every morning
The day's AI news in plain English, in your inbox at 8:30 IST. Free.
More AI news
AI unlocks Gödel's notebooks, said to contain hundreds of proved theoremsIndustry news · 13 Nov 2026Gemini 4 Hydrogen and OrchestratorIndustry news · 12 Nov 2026Gemini 4 HydrogenIndustry news · 12 Nov 2026I'm making LLMs work on small and old devicesIndustry news · 12 Nov 2026United Airlines bans AI agents from booking flights on behalf of travelersIndustry news · 11 Nov 2026This page was collected from machinelearning.apple.com and summarised automatically by AIPulsee's pipeline. See how it runs.