developer tools
The Computer-Use Agent Trains in Worlds Built to Be Checked
WildAWM synthesizes executable GUI, tool and command-line environments with solvability and verifier gates.
Summary
WildAWM synthesizes executable GUI, tool and command-line environments with solvability and verifier gates.
Training computer-use agents directly in live systems is costly and risky. WildAWM uses open-source models to synthesize controlled environments spanning graphical interfaces, tool services and command lines. Quality gates test solvability, difficulty and verifier robustness with agent runs, judge agreement and counterfactual probes. The pipeline produced 341 long-horizon environment-task pairs, some requiring more than 100 steps; a frontier model solved 60 percent. Two Qwen models trained only in these worlds improved across eight out-of-distribution benchmarks. The collection is finite, but its execution checks make generation more than prompt-written scenery.
Why it matters
WildAWM synthesizes executable GUI, tool and command-line environments with solvability and verifier gates.
Limits and context
- Two Qwen models trained only in these worlds improved across eight out-of-distribution benchmarks.
Key claims
WildAWM synthesizes executable GUI, tool and command-line environments with solvability and verifier gates.
Qualification: Two Qwen models trained only in these worlds improved across eight out-of-distribution benchmarks.
Evidence: source-2026-10-20-014
Sources
- arXiv preprint 2610.26828, original v1arXiv · primary research
Corrections
No corrections have been recorded for this story.