Web browsing on WebShop
92.6TSRFragFuse
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| FragFuseAccess Control=GuardAgent, Core LLM=GPT-4o2026.06 | 92.6 | 81 | |
| FragFuseAccess Control=LLM-AC, Core LLM=GPT-4o2026.06 | 92.5 | 93 | |
| direct queryingAccess Control=None, Core LLM=GPT-4o2026.06 | 88 | — | |
| direct queryingAccess Control=None, Core LLM=Gemini 2.5 Flash2026.06 | 88 | — | |
| direct queryingAccess Control=None, Core LLM=Average2026.06 | 83.7 | — | |
| FragFuseAccess Control=GuardAgent, Core LLM=Average2026.06 | 82.2 | 79.3 | |
| FragFuseAccess Control=GuardAgent, Core LLM=GPT-5.12026.06 | 78.6 | 84 | |
| FragFuseAccess Control=LLM-AC, Core LLM=Average2026.06 | 77.6 | 93.7 | |
| baselineAccess Control=LLM-AC, Core LLM=GPT-4o2026.06 | 77.5 | 40 | |
| baselineAccess Control=GuardAgent, Core LLM=GPT-4o2026.06 | 76.2 | 42 | |
| FragFuseAccess Control=GuardAgent, Core LLM=Gemini 2.5 Flash2026.06 | 75.3 | 73 | |
| direct queryingAccess Control=None, Core LLM=GPT-5.12026.06 | 75 | — | |
| FragFuseAccess Control=LLM-AC, Core LLM=GPT-5.12026.06 | 70.4 | 98 | |
| FragFuseAccess Control=LLM-AC, Core LLM=Gemini 2.5 Flash2026.06 | 70 | 90 | |
| baselineAccess Control=LLM-AC, Core LLM=Gemini 2.5 Flash2026.06 | 66.7 | 6 | |
| baselineAccess Control=LLM-AC, Core LLM=Average2026.06 | 54 | 30.3 | |
| baselineAccess Control=GuardAgent, Core LLM=Gemini 2.5 Flash2026.06 | 53.8 | 13 | |
| baselineAccess Control=GuardAgent, Core LLM=Average2026.06 | 51.8 | 35.3 | |
| baselineAccess Control=GuardAgent, Core LLM=GPT-5.12026.06 | 25.5 | 51 | |
| baselineAccess Control=LLM-AC, Core LLM=GPT-5.12026.06 | 17.8 | 45 |