Vision Understanding Recommendation: GPT-4o (2024-08-06)
Vision Understanding → GPT-4o (2024-08-06)
Anwendungsfallübersicht
GPT-4o (2024-08-06) by Openai is a recommended option for the Vision Understanding use case.
Empfohlene Modelle
| Rang | Anbieter | Kontextfenster | Score |
|---|
| #6 | Openai | 128K | 37/100 |
Leistungsanalyse
Benchmark-Leistung
| Benchmark | Ergebnis |
|---|
| MMLU | 87.6 |
| HUMANEVAL | 85.1 |
| GSM8K | 93.3 |
| MATH | 68.2 |
| BBH | 84.9 |
| GPQA | 62.0 |
Stärken
- MMLU score 87.6, strong knowledge reasoning.
- HumanEval 85.1, excellent code generation.
- GSM8K 93.3, robust math reasoning.
- 支持文本、图像、音频多模态输入。
Anforderungen
- 需要 openai 的 API 密钥
- 输入长度须在 128K 上下文窗口内