Phi-3-mini-4K-instruct-cpo-simpo: benchmark results
Community CPO-SimPO preference-tuned variant of Microsoft's Phi-3 Mini 4K Instruct (3.8B), reported to improve GSM8K and TruthfulQA. Provider: Syed-Hasan-8503. Released 2024-06-24. Access: Open.
Aggregate Rating 1340 ± 17, rank #1405 of 1656 rated models, from 139 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard - GPQA | 33.05 | Score | 80.7 |
| Open LLM Leaderboard - BBH | 56.82 | Score | 78.2 |
| Open Korean LLM - Ko-GPQA-Diamond | 25.25 | Normalized accuracy (%) | 75.9 |
| Open Arabic LLM - Arabic MMLU HT Abstract Algebra | 33 | Accuracy (%) | 70.4 |
| Open LLM Leaderboard - IFEval | 57.14 | Score | 69 |
| Open LLM Leaderboard - MMLU-Pro | 38.61 | Score | 68.8 |
| Open Arabic LLM - Aratrust Offensive | 84.06 | Accuracy (%) | 63.4 |
| Open LLM Leaderboard - MATH Level 5 | 15.71 | Score | 63.3 |
| Open Korean LLM - KorNAT-Harmless | 63.34 | Normalized accuracy (%) | 60.7 |
| Open Arabic LLM - Arabic MMLU HT College Mathematics | 33 | Accuracy (%) | 60.2 |
| Open Korean LLM - Ko-IFEval | 30.59 | Strict accuracy, prompt/instruction mean (%) | 57.1 |
| Open Arabic LLM - Arabic MMLU HT High School Mathematics | 30.37 | Accuracy (%) | 53.7 |
Interactive version: theaggregate.ai/model?slug=phi-3-mini-4k-instruct-cpo-simpo · How It Works · Data refreshed daily, snapshot 2026-10-20.