Cua-Bench

Make your agents better at computers

A collection of desktop and mobile computer-use tasks with a harness for evaluating and training agents, including a KiCad schematic-task suite with verifiable grading.

Recent stories

1 linked story

releasePRIMARY2026-06-15

TryCua launches Cua-Bench for KiCad; GPT-5.5 clears 6 of 25 tasks

TryCua and Snorkel opened Cua-Bench, a computer-use benchmark with 25 expert-authored KiCad tasks graded by exact netlist matches. The early results show frontier models still struggle with GUI execution, wiring completion, and self-checking, so treat benchmark wins as incomplete for real computer-use work.