Model

GPT

OpenAI · closed weights

openai.com

Coding

Coding
BenchmarkScoreMetricHarnessSourceConfidenceLink
PolyCodeEval71.7pass_raterepocoderIndependentmediumarxiv.org
PolyCodeEval68.9pass_ratehcpIndependentmediumarxiv.org
PolyCodeEval50.3pass_ratealigncoderIndependentmediumarxiv.org
PolyCodeEval48.8pass_ratellm directIndependentmediumarxiv.org

Multimodal

Multimodal
BenchmarkScoreMetricHarnessSourceConfidenceLink
ScribbleEdit0.3intention_accuracyN/AIndependentmediumarxiv.org

Reproduced by the community

Loading community runs…