Privacy Leakage Simulation on Simulation Environment
0.8Privacy Leakage (A0, D0)gpt4.1-nano vs gpt4.1-mini
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| gpt4.1-nano vs gpt4.1-miniAttack Backbone=gpt4.1-nano, Defense Backbone=gpt4.1-mini2025.08 | 0.8 | 51.3 | 3 | 21.5 | 1 | |
| gpt4.1-mini vs geminiflashAttack Backbone=gpt4.1-mini, Defense Backbone=geminiflash2025.08 | 1.5 | 39.4 | 0 | 17.1 | 2.4 | |
| qwen3-32B vs gpt4.1-miniAttack Backbone=qwen3-32B, Defense Backbone=gpt4.1-mini2025.08 | 2.9 | 54.4 | 2.6 | 20.6 | 3.4 | |
| gpt4.1-mini vs gptoss-20BAttack Backbone=gpt4.1-mini, Defense Backbone=gptoss-20B2025.08 | 3 | 48.8 | 3.6 | 11.7 | 11.6 | |
| gpt4.1-mini vs gpt4.1-miniAttack Backbone=gpt4.1-mini, Defense Backbone=gpt4.1-mini2025.08 | 3.4 | 76 | 2.5 | 42.2 | 7.1 | |
| gpt4.1-mini vs gpt4.1Attack Backbone=gpt4.1-mini, Defense Backbone=gpt4.12025.08 | 3.5 | 52.2 | 0 | 6.8 | 0 | |
| geminiflash vs gpt4.1-miniAttack Backbone=geminiflash, Defense Backbone=gpt4.1-mini2025.08 | 3.9 | 85.2 | 1 | 32.3 | 2.4 | |
| gpt4.1-mini vs qwen3-32BAttack Backbone=gpt4.1-mini, Defense Backbone=qwen3-32B2025.08 | 7 | 43.9 | 5.1 | 9.4 | 11.6 | |
| gpt4.1 vs gpt4.1-miniAttack Backbone=gpt4.1, Defense Backbone=gpt4.1-mini2025.08 | 11.9 | 79.2 | 2.8 | 38.2 | 6.7 | |
| gptoss-20B vs gpt4.1-miniAttack Backbone=gptoss-20B, Defense Backbone=gpt4.1-mini2025.08 | 12 | 65.9 | 1 | 24.5 | 1.3 | |
| gpt4.1-mini vs gpt4.1-nanoAttack Backbone=gpt4.1-mini, Defense Backbone=gpt4.1-nano2025.08 | 21.3 | 69.1 | 29.3 | 17.1 | 16.1 |