GPT-4.1
開発元 OpenAI · リリース日 2025-04-14
43.3
平均スコア
$2.00/1M
入力料金
$8.00/1M
出力料金
1.0M tokens (~524 books)
コンテキストウィンドウ
multimodal
タイプ
Tested on 22 benchmarks with 43.3% average. Top scores: HELM — WildBench (85.4%), HELM — IFEval (83.8%), MATH level 5 (83.0%).
ベンチマークスコア
| ベンチマーク | カテゴリ | スコア | Bar |
|---|---|---|---|
| HELM — WildBench | reasoning | 85.4 | |
| HELM — IFEval | language | 83.8 | |
| MATH level 5 | math | 83.0 | |
| HELM — MMLU-Pro | knowledge | 81.1 | |
| GeoBench | knowledge | 72.0 | |
| HELM — GPQA | knowledge | 65.9 | |
| Fiction.LiveBench | knowledge | 63.9 | |
| GPQA diamond | knowledge | 55.9 | |
| Aider polyglot | coding | 52.4 | |
| SWE-Bench verified | coding | 48.5 | |
| HELM — Omni-MATH | math | 47.1 | |
| CadEval | coding | 42.0 | |
| SWE-Bench Verified (Bash Only) | coding | 39.6 | |
| WeirdML | coding | 39.0 | |
| OTIS Mock AIME 2024-2025 | math | 38.3 | |
| DeepResearch Bench | knowledge | 29.3 | |
| SimpleBench | reasoning | 12.4 | |
| FrontierMath-2025-02-28-Private | math | 5.5 | |
| ARC-AGI | reasoning | 5.5 | |
| HLE | knowledge | 0.6 | |
| ARC-AGI-2 | reasoning | 0.4 | |
| FrontierMath-Tier-4-2025-07-01-Private | math | 0.1 |