NuminaMath-7B-CoT: benchmark results
Provider: Numina & Hugging Face. Released 2024-07-15. Access: Open.
Aggregate Rating 1341 ± 41, rank #7939 of 10916 rated models, from 16 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard - MATH Level 5 | 26.96 | Score | 79.4 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 41.55 | Exact match, flexible extract (%) | 76.2 |
| Open Korean LLM - Ko-GPQA-Diamond | 24.75 | Normalized accuracy (%) | 69.5 |
| Open Korean LLM - KorNAT-Helpful | 46.68 | Normalized accuracy (%) | 48.3 |
| Open Korean LLM - Ko-EQ-Bench | 0.51 | EQ-Bench score | 45.5 |
| Open Korean LLM - KorNAT-CKA | 22.69 | Normalized accuracy (%) | 40.9 |
| Open Korean LLM - Ko-IFEval | 25.99 | Strict accuracy, prompt/instruction mean (%) | 37.7 |
| Open LLM Leaderboard - MMLU-Pro | 28.68 | Score | 32.5 |
| Open LLM Leaderboard - BBH | 43.14 | Score | 29.8 |
| Open Korean LLM Leaderboard | 31.35 | Average Score (%) | 26.9 |
| Open LLM Leaderboard - IFEval | 26.89 | Score | 23.9 |
| Open LLM Leaderboard - GPQA | 26.59 | Score | 20.3 |
Interactive version: theaggregate.ai/model?slug=numinamath-7b-cot · How It Works · Data refreshed daily, snapshot 2026-11-08.