Observed arrival · 2026-10-03
Local AI, tested on one Mac
A Japanese-language benchmark compares text, translation, image, and video AI models running locally on a Mac Studio.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
- For
- Mac users comparing local AI models
- Worth noticing
- Recommendations use explicit cutoffs, such as replies under five seconds, and the site says it marks near-ties rather than forcing a winner.
Field notes
Recommendations are rule-based rather than presented as one overall ranking: the page separates options by response time, accuracy, memory use, and translation score, and says near-ties are labeled as such. It reports FLUX.2 [klein] 4B generating an image in about five seconds after the first image, and a 3.4-second Wan 2.2 TI2V-5B clip taking about 14 minutes. The homepage leaves its update date and tested-model count blank.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
○OpenPublic substance visible
✦PrettyNotable craft visible
●ProPolished or operationally mature
◎NicheUnusually specific use
One card from the complete issue