Logical Reasoning Recommendation: GPT-4o (2024-05-13)
Logical Reasoning → GPT-4o (2024-05-13)
Användningsöversikt
GPT-4o (2024-05-13) by Openai is a recommended option for the Logical Reasoning use case.
Rekommenderade modeller
| Rang | Leverantör | Kontextfönster | Score |
|---|
| #4 | Openai | 128K | 57/100 |
Prestationsanalys
Benchmarkprestanda
| Benchmark | Poäng |
|---|
| MMLU | 89.8 |
| HUMANEVAL | 83.2 |
| GSM8K | 92.6 |
| MATH | 67.1 |
| BBH | 86.7 |
| GPQA | 58.8 |
Styrkor
- MMLU score 89.8, strong knowledge reasoning.
- HumanEval 83.2, excellent code generation.
- GSM8K 92.6, robust math reasoning.
- 支持文本、图像、音频多模态输入。
Krav
- 需要 openai 的 API 密钥
- 输入长度须在 128K 上下文窗口内