Apple ML Research: Muse Spark Open with MLX on Mac Studio

tony_cannistra

Not exactly your average "desk", but still neat. In this post, we show how to run the latest frontier-class model — Muse Spark Open, with nearly a trillion parameters — right from your desk, using the latest version of MLX, released today. On a cluster of four Mac Studios, Muse Spark Open generates over 36 tokens per second (see Figure 1), well above average human reading speed. Beginning on a single Mac Studio (up to 512GB of unified memory), then scaling out across multiple Macs with the RDMA over Thunderbolt driver, below we share an overview of the range of models you can now run on Mac Studio.

janderson215

Hilarious how they managed to put out an entire press release about Meta’s technology without mention Meta other than what is in the code references.

peri-cl

I think they're sandbagging Qwen pretty heavily by benchmarking Meta's model at 4-bit, against Qwen only at native BF16 (and nothing else shown). That's not a quant most users would choose. I get that this is cross-brand marketing for Apple's partner, Meta; but really Muse Spark isn't even in the top *ten* models I'd consider if I had a 512 GB M5 in my paws. Or a cluster of them.

butterlesstoast

I -- love -- an article -- full of -- em -- dashes.

isodev

Preparing to upsell those very expensive Macs coming later this month, are we.

ks2048

So, Meta will spend millions/billions on building open-weight models and then people will send Apple $20,000 to run them. Seems like a good deal for Apple.

_pdp_

I am super excited about this. While right now Mac Studio is not a budget device and it certainly costs more than the subsidised subscription, the promise that one day we can run these models on commodity hardware is exciting.

comandillos

A 800B model that performs way worse than other 300B ones. Well...

magnusokik

It's a good article which showcases direction of travel, i am sure mac studio will get more and more powerful in coming generations but for now it's way to expensive for local setup.

liuliu

FWIW, single M5 Ultra with 256GiB runs DeepSeek 4.1 Flash 0910 just fine (~3200 tok/s prefill, ~45 tok/s). M5 Max w/ 128GiB runs GLM 5.3 Flash also fine. I also benchmarked both on quality benchmarks (MathArena Apex / GPQA Diamond etc) to make sure the quality is what you get from the official numbers.

throwworhtthrow

> peak power consumption is about 200W for the Mac Studio with M3 Ultra, and 300W for the Mac Studio with M5 Ultra — a fraction of the power used by a common toaster . (Emphasis added.)

AndrewKemendo

Anyone know why they went with Meta models?

nico

Very cool Has anyone adapted the open Muse model to work like a decision model? Sort of like Laya but based on muse spark

mrboatshoepants

That first bar chart showing between 15-36 tokens/second is for a 8,192 token context. Thats not nearly enough context to be useful these days

macinjosh

Many thought Apple should be competing with the frontier labs while they see themselves as competing with Nvidia.