ResearchBenchmarksWeb browsing on Custom TasksFollow57.7ScoreGA49.69251.77153.8555.929Apr 18, 2026Evaluation ResultsMethodMethodLinksScoreAverage Token Usage (M)GATasks=22, Evaluation=H...Tasks=22, Evaluation=Human + LLM, Backbone model=Claude Opus 4.62026.0457.70.26OpenClawTasks=22, Evaluation=H...Tasks=22, Evaluation=Human + LLM, Backbone model=Claude Opus 4.62026.04500.76