Phi-3 (7B): benchmark results
Provider: Microsoft. Access: Open.
Aggregate Rating 1503 ± 24, rank #2292 of 10916 rated models, from 85 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| HELM MMLU - Us Foreign Policy | 95 | EM | 83.3 |
| HELM MMLU - High School European History | 87.27 | EM | 78.2 |
| HELM Lite - MATH Level 1 - Algebra | 93.33 | Equivalent (CoT) (%) | 77.8 |
| HELM MMLU - Management | 90.29 | EM | 76.3 |
| HELM MMLU - Moral Scenarios | 71.06 | EM | 70.5 |
| HELM MMLU - High School Mathematics | 49.63 | EM | 69.2 |
| HELM Lite - MATH Level 1 - Geometry | 60.53 | Equivalent (CoT) (%) | 68.3 |
| HELM MMLU - College Physics | 55.88 | EM | 67.9 |
| HELM NarrativeQA | 75.4 | F1 (%) | 67.8 |
| HELM Lite - MATH Level 1 - Counting and Probability | 76.92 | Equivalent (CoT) (%) | 66.7 |
| HELM MMLU - Machine Learning | 65.18 | EM | 66 |
| HELM MMLU - High School Microeconomics | 89.92 | EM | 64.7 |
Interactive version: theaggregate.ai/model?slug=phi-3-7b · How It Works · Data refreshed daily, snapshot 2026-11-08.