Agent World Model in the Wild: Computer-Use Agent Environment Synthesis
arXiv cs.AI · 20 Oct 2026, 03:55 AEST
Agent learning is evolving towards interaction with the world rather than relying solely on human data. The proposed Agent World Model in the Wild (WildAWM) offers a scalable method for creating environments that train general computer-use agents.
Key facts
- We therefore propose Agent World Model in the Wild (WildAWM), a scalable pipeline for synthesizing controllable environments and tasks for training general computer-use agents (CUAs).
- WildAWM is entirely powered by open-source models, demonstrating that high-quality computer-use environments can be generated without proprietary frontier models.
- The resulting pipeline synthesizes 341 long-horizon environment-task pairs that can require over 100 agent steps to solve and remain challenging even for frontier models.
- Both models achieve substantial gains across 8 out-of-distribution benchmarks spanning web, claw, and computer-use agents, demonstrating that learning in WildAWM's synthesized worlds generalizes broadly to unseen computer-use tasks.
Read aloud
Agent learning is evolving towards interaction with the world rather than relying solely on human data. The proposed Agent World Model in the Wild WildAWM offers a scalable method for creating environments that train general computer-use agents. We therefore propose Agent World Model in the Wild WildAWM, a scalable pipeline for synthesizing controllable environments and tasks for training general computer-use agents CUAs. WildAWM is entirely powered by open-source models, demonstrating that high-quality computer-use environments can be generated without proprietary frontier models. The resulting pipeline synthesizes three hundred forty one long-horizon environment-task pairs that can require over one hundred agent steps to solve and remain challenging even for frontier models. Both models achieve substantial gains across eight out-of-distribution benchmarks spanning web, claw, and computer-use agents, demonstrating that learning in WildAWM's synthesized worlds generalizes broadly to unseen computer-use tasks.
Sources in this cluster
- Agent World Model in the Wild: Computer-Use Agent Environment Synthesis
arXiv cs.AI · 20 Oct 2026, 03:55 AEST