GPT-5.1
来自 OpenAI · 发布于 2025-11-13
49.6
平均分
$1.25/1M
输入价格
$10.00/1M
输出价格
400K tokens (~200 books)
上下文窗口
multimodal
类型
Tested on 24 benchmarks with 49.6% average. Top scores: Chatbot Arena Elo — Overall (1438.5%), Chatbot Arena Elo — Coding (1338.8%), HELM — IFEval (93.5%).
基准测试分数
| 基准测试 | 类别 | 分数 | Bar |
|---|---|---|---|
| Chatbot Arena Elo — Overall | arena | 1438.5 | |
| Chatbot Arena Elo — Coding | arena | 1338.8 | |
| HELM — IFEval | language | 93.5 | |
| OTIS Mock AIME 2024-2025 | math | 88.6 | |
| HELM — WildBench | reasoning | 86.3 | |
| GPQA diamond | knowledge | 83.5 | |
| ARC-AGI | reasoning | 72.8 | |
| SWE-Bench verified | coding | 68.0 | |
| SWE-Bench Verified (Bash Only) | coding | 66.0 | |
| WeirdML | coding | 60.8 | |
| HELM — MMLU-Pro | knowledge | 57.9 | |
| SimpleQA Verified | knowledge | 48.9 | |
| Terminal Bench | coding | 47.6 | |
| HELM — Omni-MATH | math | 46.4 | |
| HELM — GPQA | knowledge | 44.2 | |
| SimpleBench | reasoning | 43.8 | |
| VPCT | knowledge | 38.0 | |
| Chess Puzzles | knowledge | 32.0 | |
| FrontierMath-2025-02-28-Private | math | 31.0 | |
| HLE | knowledge | 19.8 | |
| ARC-AGI-2 | reasoning | 17.6 | |
| APEX-Agents | agentic | 17.5 | |
| GSO-Bench | coding | 13.7 | |
| FrontierMath-Tier-4-2025-07-01-Private | math | 12.5 |