ruGPT-3.5: benchmark results

Provider: Other. Access: Open.

Unified ELO 1182 ± 39, rank #1470 of 1537 rated models, from 31 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
MERA Original Paper20.8Total Score (%; mean of 17 problem-solving and exam tasks, M72.2
MERA Original Paper - SimpleAr2.9Exact Match (%; 5-shot, greedy generation)72.2
MERA Original Paper - ruMultiAr2.5Exact Match (%; 5-shot, greedy generation)72.2
MERA Original Paper - ruModAr0.1Exact Match (%; greedy generation)58.3
MERA - LCS13.2Accuracy (%)53.1
MERA Original Paper - MathLogicQA25.8Accuracy (%; 5-shot, log-likelihood)47.2
MERA - CheGeKa14.47F1 (%)47.1
MERA Original Paper - ruMMLU24.6Accuracy (%; 5-shot, log-likelihood)38.9
MERA Original Paper - ruWorldTree24.6Accuracy (%; 5-shot, log-likelihood)38.9
MERA - ruCodeEval0.24pass@1 (%)23.2
MERA - USE8.24Grade, normalized (%)22.2
MERA - ruHumanEval1.04pass@1 (%)21.9

Interactive version: theaggregate.ai/model?slug=rugpt-3-5 · How It Works · Data refreshed daily, snapshot 2026-09-25.