Model

Sonnet

Anthropic · closed weights

anthropic.com

Coding

Coding
BenchmarkScoreMetricHarnessSourceConfidenceLink
PolyCodeEval59.4pass_ratealigncoderIndependentmediumarxiv.org
PolyCodeEval58.7pass_raterepocoderIndependentmediumarxiv.org
PolyCodeEval43.2pass_ratehcpIndependentmediumarxiv.org
PolyCodeEval40.4pass_ratellm directIndependentmediumarxiv.org

Safety & Alignment

Safety & Alignment
BenchmarkScoreMetricHarnessSourceConfidenceLink
RT-Safe2.8safe_success_rateN/AIndependentmediumarxiv.org

Reproduced by the community

Loading community runs…