ruGPT3-large: benchmark results

Provider: Other. Access: Open.

Unified ELO 1164 ± 39, rank #1484 of 1537 rated models, from 31 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
MERA Original Paper - ruModAr0.1Exact Match (%; greedy generation)58.3
MERA Original Paper - SimpleAr0.4Exact Match (%; 5-shot, greedy generation)50
MERA - LCS12Accuracy (%)44.7
MERA Original Paper - ruMultiAr0.7Exact Match (%; 5-shot, greedy generation)44.4
MERA Original Paper - ruMMLU24.5Accuracy (%; 5-shot, log-likelihood)33.3
MERA Original Paper - MathLogicQA25.1Accuracy (%; 5-shot, log-likelihood)27.8
MERA - USE8.73Grade, normalized (%)25.8
MERA Original Paper19.3Total Score (%; mean of 17 problem-solving and exam tasks, M19.4
MERA - ruHateSpeech54.34Accuracy (%)13.2
MERA - RWSD46.92Accuracy (%)12.4
MERA Original Paper - ruWorldTree23.2Accuracy (%; 5-shot, log-likelihood)11.1
MERA - MultiQ13.55F1 (%)11

Interactive version: theaggregate.ai/model?slug=rugpt3-large · How It Works · Data refreshed daily, snapshot 2026-09-25.