ResearchBenchmarksSafety Verdict Classification on LIBEROFollow92.5P10 Success RateDILLO-4B88.44489.49790.5591.603Mar 24, 2026Evaluation ResultsMethodMethodLinksP10 Success RateP90 Success RateGoal Success RateObject Success RateSpatial Success RateDILLO-4BModel Variant=4BModel Variant=4B2026.0392.585.787.581.286.7DILLO-1BModel Variant=1BModel Variant=1B2026.0388.686.98079.877.8