GoKawiilMeta releases 800B-parameter Muse Spark Open model, tuned for Mac Studio clusters
Meta has open-sourced Muse Spark Open, an 800-billion-parameter model it calls the most capable US open-weight model yet. Running it locally requires roughly 1TB of unified memory, achievable by pairing two of Apple's upcoming M5 Ultra Mac Studios with 512GB each, or four 256GB units costing about $38,000. The weights are now downloadable from Hugging Face, and Meta says the model was built in collaboration with Apple to run on its new hardware.
GoKawiil's interpretation of the reporting above, not reported fact.
Meta's own benchmarks claim the model beats Chinese rivals GLM 5.3 and Kimi K3 on agentic and visual reasoning tasks, positioning it as a flagship example of Meta's push for US leadership in open-weight AI. The steep hardware requirement suggests this release targets well-funded enterprises rather than individual developers, potentially driving institutional demand for high-end Mac Studio clusters as an alternative to cloud API costs. Apple's involvement also hints at a deepening hardware-software partnership aimed at making consumer-grade (if expensive) machines viable for running frontier-scale models locally.
- Muse Spark Open is an 800B-parameter open-weight model requiring about 1TB of RAM to run locally.
- It can run on two 512GB M5 Ultra Mac Studios or four 256GB units costing around $38,000.
- Meta claims it outperforms Chinese models GLM 5.3 and Kimi K3 on agentic and visual reasoning benchmarks.
Apple Mac Studio (M5 Ultra) — This is literally the machine Meta built Muse Spark Open around—its unified memory architecture is what makes running an 800B parameter model locally even conceivable. If you want to experiment with cutting-edge open-weight AI on Apple Silicon, the Mac Studio is the hardware purpose-fit for the job.
See Apple Mac Studio (M5 Ultra) on Amazon → Affiliate link — we may earn a commission on purchases, at no extra cost to you. Product picked by AI based on this article; it is not a tested recommendation.Source: engadget.com — Karissa Bell, 2026-10-15
Published there as: “Meta built a huge open-weight AI model for Macs, but you'll need 1TB of RAM to run it”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.