ResearchBenchmarksOut-of-Distribution Evaluation on GAIAFollow32.5Avg@4CARL26.98828.41929.8531.281Dec 4, 2025Evaluation ResultsMethodMethodLinksAvg@4Pass@4CARLModel Size=4B reasonin...Model Size=4B reasoning model, Action Budget=32 actions, Number of Seeds=42025.1232.554.4GRPOModel Size=4B reasonin...Model Size=4B reasoning model, Action Budget=32 actions, Number of Seeds=42025.1230.649.5Zero-ShotModel Size=4B reasonin...Model Size=4B reasoning model, Action Budget=32 actions, Number of Seeds=42025.1227.244.7