Reproduce “OpenGCUA: Toward a Fully Open Stack for Generalist Computer Use” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Hanzhang Zhou, Long Li, Quyu Kong, Lei Wang, Panrong Tong, Long Chen, Tianyu Xia, Gongjie Zhang and 7 more
OpenGCUA lets a computer-use model coordinate interface actions and work organization through executable programs. Programs can inspect intermediate results, choose subsequent operations, request screenshots when needed, and track unfinished work, rather than committing to fixed action batches. On OSWorld-v2’s 82-task offline subset, the full configuration raised mean partial scores by 19.6 points and reduced model-call steps by 53%, compared with a batched GUI-only baseline.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.