Multimodal Safety Evaluation on Image input safety evaluation set
98.6Hate Safety Accgpt-5-thinking-nano
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| gpt-5-thinking-nano2025.12 | 98.6 | 97.3 | 98.6 | 98.6 | 93.9 | 96.3 | |
| gpt-5-main-mini2025.12 | 98.4 | 98.4 | 98.2 | 99.5 | 99.4 | 99.8 | |
| gpt-5-thinking-mini2025.12 | 97.1 | 98.2 | 98.6 | 98.6 | 98.7 | 99.2 | |
| OpenAI o4-mini2025.12 | 92.7 | 95 | 95.6 | 93.9 | 92.7 | 97.8 |