Model

Llama-3.1-8B-Instruct

Meta · open weights

ai.meta.com

Knowledge & Reasoning

Knowledge & Reasoning
BenchmarkScoreMetricHarnessSourceConfidenceLink
Balrog15.1average_progressN/AIndependentmediumbalrogai.com
BLiMPPI92.7accuracyN/AIndependentmediumarxiv.org
BoolQ82.8scoreN/AIndependentmediumarxiv.org
CritPt0.0accuracyN/AIndependentlowepoch.ai
GeopoliticalQA2.3overallN/AIndependentmediumarxiv.org
GPQA Diamond27.0accuracyN/AIndependenthighepoch.ai
LOGIC0.8f1N/AIndependentmediumarxiv.org
MMLU73.0average_(%)N/AIndependentmediumarxiv.org
MMLU56.1exact_matchN/AIndependentmediumcrfm.stanford.edu
MS-Exam-Gen73.1accuracyN/AIndependentmediumarxiv.org
PIQA81.2scoreN/AIndependentmediumarxiv.org
ViLegalExpert72.1scoreN/AIndependentmediumarxiv.org

Math

Math
BenchmarkScoreMetricHarnessSourceConfidenceLink
GSM8K82.4exact_matchN/AIndependentmediumarxiv.org

Coding

Coding
BenchmarkScoreMetricHarnessSourceConfidenceLink
AlgoREval31.4accuracyN/AIndependentmediumarxiv.org
PatchBench31.2hncsN/AIndependentmediumarxiv.org
SciCode13.2scoreN/AIndependentmediumartificialanalysis.ai

Agentic / Tool Use

Agentic / Tool Use
BenchmarkScoreMetricHarnessSourceConfidenceLink
MASBench26.1scoreN/AIndependentmediumarxiv.org
NAQD-Env0.6withdrawal_recallnaqd-envIndependentmediumarxiv.org
SmartSalesBench24.2crN/AIndependentmediumarxiv.org
TALK-Dem66.7success_rateN/AIndependentmediumarxiv.org

Multimodal

Multimodal
BenchmarkScoreMetricHarnessSourceConfidenceLink
VisualNoiseQA36.4accuracyreactIndependentmediumarxiv.org

Instruction Following

Instruction Following
BenchmarkScoreMetricHarnessSourceConfidenceLink
CreativePreferences0.7aucN/AIndependentmediumarxiv.org
Intent-Eval91.7accuracyN/AIndependentmediumarxiv.org

Safety & Alignment

Safety & Alignment
BenchmarkScoreMetricHarnessSourceConfidenceLink
ObligationBench0.6recallN/AIndependentmediumarxiv.org
OMIT0.5obrN/AIndependentmediumarxiv.org
PFaithBench48.7overansN/AIndependentmediumarxiv.org

Reproduced by the community

Loading community runs…