GPT-4.1 (2025-04-14): benchmark results
April 14, 2025 GPT-4.1 snapshot, tracked when sources report the dated API model. Provider: OpenAI. Released 2025-04-14. Access: API.
Aggregate Rating 1678 ± 9, rank #250 of 1656 rated models, from 1337 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| HELM Long Context - InfiniteBench En.MC | 97 | EM | 100 |
| HELM Long Context - RULER HotPotQA | 70 | RULER String Match | 100 |
| HELM Long Context - RULER SQuAD | 88 | RULER String Match | 100 |
| HELM Robo-Reward-Bench - Robo Reward Bench Discrete | 1.4 | Absolute error | 100 |
| HELM Robo-Reward-Bench - Robo Reward Bench Preference Ranking | 67.5 | EM | 100 |
| HELM VHELM - Exams V - Bulgarian (Natural Science) | 46.3 | Prefix Quasi-Exact Match (%) | 100 |
| MIA-Bench | 94.55 | Overall Score | 100 |
| MedAgentGym | 70.15 | Average Score (self-reported) | 100 |
| Open LMM Reasoning - DynaMath - Subject-statistics; Avg | 79.6 | Accuracy (%) | 100 |
| Open LMM Reasoning - MathVision - analytic geometry | 58.3 | Accuracy (%) | 100 |
| Open LMM Reasoning - MathVision - arithmetic | 66.4 | Accuracy (%) | 100 |
| OpenCompass Knowledge - Humanities - Chinese (CompassBench 2507) | 92 | Score (%) | 100 |
Interactive version: theaggregate.ai/model?slug=gpt-4-1-2025-04-14 · How It Works · Data refreshed daily, snapshot 2026-10-20.