What changed
Researchers tested longer context, longer trajectories, task decomposition and parallel attempts for selected Qwen3-VL, UI-TARS and OpenCUA models on OSWorld. Extra context saturated, long runs prolonged some wrong paths, decomposition added overhead and parallel attempts were expensive.
Why it matters
More steps and context can spend additional compute on the same mistaken plan. Reliability needs failure detection and selective escalation, not an unrestricted execution budget.
What to do
Detect stalled loops, repeated actions and premature success claims. Escalate context, steps or parallel attempts only when a failure signal suggests another try can help.
Caveat
The study covers selected locally hosted models on one desktop benchmark. Its findings should not be generalized to every frontier, hosted or domain-specific agent.