Clipboard GUI Leaderboard
60+ models, one GUI prompt. Ranked by what actually shipped: does it run, does it look right. Click a header to sort, a chip to filter, a name for its page. For the 36-task hidden-test suite, see the AHK-Eval leaderboard.
Every model, plotted: code score (x) vs. what actually shipped (y)
All
Runs
Crashes
Won’t parse
↻ Recovered on fallback
| # | Model | runs | parse | code | visual | OOP |
|---|