Model

Qwen3 4B

Alibaba · open weights

qwenlm.github.io

Knowledge & Reasoning

Knowledge & Reasoning
BenchmarkScoreMetricHarnessSourceConfidenceLink
GPQA Diamond52.3accuracyN/AIndependenthighepoch.ai
OUC-Benchmark53.0f1N/AIndependentmediumarxiv.org
PhysicsMate27.6accuracyN/AIndependentmediumarxiv.org

Agentic / Tool Use

Agentic / Tool Use
BenchmarkScoreMetricHarnessSourceConfidenceLink
MAGMA-Bench10.6success_rateN/AIndependentmediumarxiv.org
ToolRACERBench75.9accuracyN/AIndependentmediumarxiv.org

Safety & Alignment

Safety & Alignment
BenchmarkScoreMetricHarnessSourceConfidenceLink
Guise92.9attack_success_rateN/AIndependentmediumarxiv.org
MediateRec8.5resc.N/AIndependentmediumarxiv.org
PFaithBench27.3overansN/AIndependentmediumarxiv.org

Reproduced by the community

Loading community runs…