Phi-4 (Publisher: unsloth): benchmark results
Provider: Microsoft. Access: Open.
Aggregate Rating 1606 ± 52, rank #957 of 10916 rated models, from 12 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard - BBH | 68.86 | Score | 97.3 |
| Open LLM Leaderboard - MATH Level 5 | 50 | Score | 96.9 |
| Open LLM Leaderboard - MMLU-Pro | 53.78 | Score | 94.8 |
| Open LLM Leaderboard - GPQA | 33.64 | Score | 83.1 |
| Open LLM Leaderboard - IFEval | 68.82 | Score | 81.8 |
| JP-TL-Bench - Japanese to English | 8.72 | LT Score (0-10; anchored pairwise Bradley-Terry vs 20 Base S | 65.1 |
| JP-TL-Bench - Japanese to English - Easy | 8.61 | LT Score (0-10; Easy items only; anchored pairwise Bradley-T | 63.5 |
| JP-TL-Bench - Japanese to English - Hard | 8.9 | LT Score (0-10; Hard items only; anchored pairwise Bradley-T | 63.5 |
| JP-TL-Bench - English to Japanese - Easy | 8.87 | LT Score (0-10; Easy items only; anchored pairwise Bradley-T | 60.3 |
| JP-TL-Bench - English to Japanese | 8.7 | LT Score (0-10; anchored pairwise Bradley-Terry vs 20 Base S | 58.7 |
| JP-TL-Bench - English to Japanese - Hard | 8.75 | LT Score (0-10; Hard items only; anchored pairwise Bradley-T | 55.6 |
| Open LLM Leaderboard - MuSR | 41.14 | Score | 51.6 |
Interactive version: theaggregate.ai/model?slug=phi-4-publisher-unsloth · How It Works · Data refreshed daily, snapshot 2026-11-08.