Math Reasoning Recommendation: GPT-4o
Math Reasoning → GPT-4o
Aperçu du cas d’usage
GPT-4o by Openai is a recommended option for the Math Reasoning use case.
Modèles recommandés
| Rang | Fournisseur | Fenêtre de contexte | Score |
|---|
| #1 | Openai | 128K | 57/100 |
| Benchmark | Score |
|---|
| MMLU | 88.7 |
| HUMANEVAL | 90.2 |
| GSM8K | 95.8 |
| MATH | 76.6 |
| BBH | 83.1 |
Forces
- MMLU score 88.7, strong knowledge reasoning.
- HumanEval 90.2, excellent code generation.
- GSM8K 95.8, robust math reasoning.
- 支持文本、图像、音频多模态输入。
Exigences
- 需要 openai 的 API 密钥
- 输入长度须在 128K 上下文窗口内