The trick: Lab Not Field
Meta's 816-billion-parameter model runs on two Mac Studios.
Each one needs 512GB of memory, a model MacRumors says Apple has not released yet.
Meta's 816B Muse Spark Open runs locally on paired Mac Studios.
Before you read on. Your call?
TRUE, BUT
15.74
MacRumors says it needs two 512GB M5 Ultra Mac Studios, and that Apple has not released the 512GB model yet. Apple ran it in 4-bit precision. Apple's over-36 tokens a second came from four Mac Studios; Let's Data Science reports 15.74 for a two-Mac pipeline.
There’s more to this story.
Membership opens the full investigation, the strongest counterargument and what to do with what you’ve learned.
30 days free, then A$89 a year. Payment card required. Cancel before the trial ends to avoid the first charge.
Review my 30-day free trial →- At sign-up: 30 days free. Payment card required. One introductory trial per customer.
- Your trial ends 30 days after you start. Unless you cancel before then in Account → Manage subscription, we charge A$89 for the first year. It renews automatically at A$89/yr until cancelled.
- You can ask for a full refund within 14 days after any annual payment, renewals included, with no reason needed, by emailing hello@bskiller.com. This voluntary refund does not limit your rights under the Australian Consumer Law.
By starting your trial, you agree to the Terms.
BS Killer is published by Inferno Tech Pty Ltd, ABN 27 647 413 474.
Get tomorrow’s check free by email.
Couldn't check your access. That's on us.
The trick has a name
We call it Lab Not Field: it works in the test conditions, not the deployed ones. You'll see it again. Learn to spot it →
Receipts
- Supports macrumors.com:
most capable U.S. open-weight model to date
- Supports macrumors.com:
two 512GB Apple M5 Ultra Mac Studios running the latest MLX framework
- Refutes macrumors.com:
as of yet, but it's supposed to be coming later this month
- Context macrumors.com:
It won't be cheap, because the 256GB
- Context machinelearning.apple.com:
On a cluster of four Mac Studios, Muse Spark Open generates over 36 tokens per second
- Context machinelearning.apple.com:
Muse Spark Open in 4-bit (NVFP4) precision
- Refutes letsdatascience.com:
15.74 generated tokens per second for a two-Mac pipeline setup
- Context letsdatascience.com:
Meta's claim that this is the most capable U.S. open-weight model is its own assessment
- Context huggingface.co:
requires a GPU with at least 700GB of VRAM for NVFP4 and 2TB for BF16
Open the Receipts Pack → What each source proves, every figure traced, and what would change our verdict.
