23.10.2026·00:01:23

You Can Run Muse Spark on Your Mac... for $20,000

Meta has released Muse Spark Open, the open-weight version of Muse Spark. It's an 800GB model that runs on two Apple M5 Ultra Mac Studios with 512GB of RAM. Sam goes through the naming, the local benchmarks, MLX support and the 4-bit version on Hugging Face. — TIMESTAMPS: 00:00 An 800GB model on two Mac Studios 00:11 Muse Spark Open is here 00:45 Local benchmarks and the Hugging Face versions 01:07 Outro Would you run an 800GB model at home? Taken from Cmd+ AI Ep 70, our weekly live show covering AI, design and dev. Catch us live every Friday and join in the banter. — Unlock the full potential of your online presence with Kabarza and Samuel—experts in web design and development (respectively), powered by cutting-edge AI solutions. We blend creative design with advanced tech to deliver smart, high-impact websites that stand out. Ready to elevate your business? Contact us today and see what AI-driven innovation can do for you! LINKS & RESOURCES: Website: https://cmdaishow.com Check out Kabarza's amazing work: https://kabarza.com Visit Samuel's website for more: https://samuelgregory.co.uk 📷 Follow on Instagram: https://www.instagram.com/cmdaishow — HASHTAGS: #Meta #Muse #OpenSource #LocalAI #CmdAI #ai #podcast #aidesign #aidevelopment #vibecoding #webdesign #webdevelopment #ainews #webnews #designnews #devnews

Transcript

And you can run it on two Apple M5 Ultra Mac Studios with 512 gig of RAM cuz it's a 800 gig model. Command AI On the subject of Muse, you can now run Muse on your Mac in like a con I think it was a collaboration with Apple. You can now run Muse open. Muse Spark open. All of these The naming's getting very confusing in my eyes. Muse Spark open is the open weight version of Muse Spark, which is the lower model. And you can run it on two Apple M5 Ultra Mac Studios with 512 gig of RAM cuz it's a 800 gig model. So this is quite cool. Again, benchmarks is a local model. You can't expect too much. Optimized for Apple Silicon. Running the latest version of MLA. You can get it on Hugging Face here. There's two versions. There is a quantized 4-bit version, which is 484 billion parameters. Two Mac Studios is what it takes to run this bloody model. So That's crazy. They're still contributing to the open source stuff, which is cool. This is part of a larger conversation on my show, Command AI, which we stream live every single week. We discuss the news and all things related to AI in the world of design and web. Catch us next week and join in the banter. Command AI