Model

Gemini

Google

deepmind.google

Knowledge & Reasoning

Knowledge & Reasoning
BenchmarkScoreMetricHarnessSourceConfidenceLink
SHAD0.2anomaly_score_contribution_and_llm_interpretation_qualityN/AIndependentmediumarxiv.org

Agentic / Tool Use

Agentic / Tool Use
BenchmarkScoreMetricHarnessSourceConfidenceLink
Talk2Agent40.8success_rateN/AIndependentmediumarxiv.org

Multimodal

Multimodal
BenchmarkScoreMetricHarnessSourceConfidenceLink
ChartDensity-Bench5.5maeN/AIndependentmediumarxiv.org
SocialNav-SUB0.5paN/AIndependentmediumarxiv.org
VisAudit68.0recovery_rateN/AIndependentmediumarxiv.org
VNBench66.7accuracyN/AIndependentmediumarxiv.org
Winoground30.8text_scoreN/AIndependentmediumarxiv.org

Safety & Alignment

Safety & Alignment
BenchmarkScoreMetricHarnessSourceConfidenceLink
RT-Safe0.0safe_success_rateN/AIndependentmediumarxiv.org

Reproduced by the community

Loading community runs…