GPT-4.1 Mini
제공 OpenAI · 출시일 2025-04-14
44.5
평균 점수
$0.40/1M
입력 가격
$1.60/1M
출력 가격
1.0M tokens (~524 books)
컨텍스트 윈도우
multimodal
유형
Tested on 16 benchmarks with 44.5% average. Top scores: HELM — IFEval (90.4%), MATH level 5 (87.3%), HELM — WildBench (83.8%).
벤치마크 점수
| 벤치마크 | 카테고리 | 점수 | Bar |
|---|---|---|---|
| HELM — IFEval | language | 90.4 | |
| MATH level 5 | math | 87.3 | |
| HELM — WildBench | reasoning | 83.8 | |
| HELM — MMLU-Pro | knowledge | 78.3 | |
| HELM — GPQA | knowledge | 61.4 | |
| GPQA diamond | knowledge | 54.5 | |
| HELM — Omni-MATH | math | 49.1 | |
| OTIS Mock AIME 2024-2025 | math | 44.7 | |
| Fiction.LiveBench | knowledge | 44.4 | |
| WeirdML | coding | 37.6 | |
| Aider polyglot | coding | 32.4 | |
| SWE-Bench Verified (Bash Only) | coding | 23.9 | |
| CadEval | coding | 16.0 | |
| FrontierMath-2025-02-28-Private | math | 4.5 | |
| ARC-AGI | reasoning | 3.5 | |
| ARC-AGI-2 | reasoning | 0.1 |