GPT-4 (older v0314)
来自 OpenAI · 发布于 2023-05-28
55.0
平均分
$30.00/1M
输入价格
$60.00/1M
输出价格
8K tokens (~4 books)
上下文窗口
text
类型
Tested on 7 benchmarks with 55.0% average. Top scores: Chatbot Arena Elo — Overall (1285.8%), GSM8K (92.0%), MMLU (81.9%).
基准测试分数
| 基准测试 | 类别 | 分数 | Bar |
|---|---|---|---|
| Chatbot Arena Elo — Overall | arena | 1285.8 | |
| GSM8K | math | 92.0 | |
| MMLU | knowledge | 81.9 | |
| Winogrande | knowledge | 75.0 | |
| Aider — Code Editing | coding | 66.2 | |
| GPQA diamond | knowledge | 14.3 | |
| OTIS Mock AIME 2024-2025 | math | 0.5 |