About this site
We test local AI on Macs. When we compare a cloud tool, we identify where it processes the input.
Recent tests and updates
- Published the seven task entrances and test plans
- Text model measurement or update · Hy-Memory-30B-A3B
- Text model measurement or update · MiniCPM5-2B
- Text model measurement or update · muse-glimmer:30b
How we test
Compare runs with the same task and settings. Results can change with the chip, memory, model version, input and app. A model that runs can still need review and manual fixes.