About

Wide bench shot of four benchmark machines on a home-office desk, with a child's drawing of a flower in the bottom-right corner.

This site publishes honest benchmarks for personal AI on consumer hardware — gaming laptops, base-model Macs, cheap VPSes — so people deciding what to run can know what to expect before they spend the time.

Each benchmark publishes its raw data, its methodology, and the git SHA of the harness that produced it. The point isn’t to sell anything; it’s to save the next person the same dead ends.

Built and run by Sloba Margetic. Contributions help a lot — if the work is useful to you, please visit my GitHub sponsors page at github.com/sponsors/slobodanmargetic988.

The wider Weeyuga project lives at weeyuga.com; this site is the benchmarks corner of it — for engineers who want numbers.

Public archive on GitHub →