Model

llama-4 Scout

Meta · open weights

ai.meta.com

Coding

Coding
BenchmarkScoreMetricHarnessSourceConfidenceLink
Hob-VL50.0accuracyN/AIndependentmediumarxiv.org
PatchBench18.0hncsN/AIndependentmediumarxiv.org
Web-Bench5.0pass@1N/AIndependentmediumarxiv.org

Agentic / Tool Use

Agentic / Tool Use
BenchmarkScoreMetricHarnessSourceConfidenceLink
GeoNatureAgent (GNA)26.9accuracyN/AIndependentmediumarxiv.org

Safety & Alignment

Safety & Alignment
BenchmarkScoreMetricHarnessSourceConfidenceLink
RH-Detect0.9aurocN/AIndependentmediumarxiv.org

Reproduced by the community

Loading community runs…