Web Navigation on Mind2Web (AVG)
66.8Execution Accuracy (EA)FluxMem
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| FluxMemBase Model=GPT-4.1-mini, Setting=Filtered (manual removal of non-golden HTML elements)2026.05 | 66.8 | 77.9 | 56.6 | 8.6 | |
| FluxMemBase Model=Gemini-2.5-flash, Setting=Filtered (manual removal of non-golden HTML elements)2026.05 | 66.2 | 73.6 | 56.5 | 5.9 | |
| FluxMemBase Model=Gemini-2.5-flash, Setting=Realistic (without any pre-filtering)2026.05 | 59.2 | 71.8 | 56.1 | 6.2 | |
| FluxMemBase Model=GPT-4.1-mini, Setting=Realistic (without any pre-filtering)2026.05 | 57.6 | 71.7 | 53.4 | 5.5 | |
| AWMBase Model=Gemini-2.5-flash, Setting=Filtered (manual removal of non-golden HTML elements)2026.05 | 56.3 | 71.2 | 51.3 | 4.5 | |
| AWMBase Model=GPT-4.1-mini, Setting=Filtered (manual removal of non-golden HTML elements)2026.05 | 54.8 | 73.1 | 47.4 | 3.7 | |
| AWMBase Model=Gemini-2.5-flash, Setting=Realistic (without any pre-filtering)2026.05 | 44.7 | 64.8 | 40.8 | 2.5 | |
| ReasoningBankBase Model=Gemini-2.5-flash, Setting=Filtered (manual removal of non-golden HTML elements)2026.05 | 43.2 | 46.4 | 37.8 | 2.4 | |
| AWMBase Model=GPT-4.1-mini, Setting=Realistic (without any pre-filtering)2026.05 | 32.2 | 60.4 | 26.3 | 1.5 | |
| No MemoryBase Model=GPT-4.1-mini, Setting=Realistic (without any pre-filtering)2026.05 | 31.3 | 58.1 | 25.2 | 1.1 | |
| No MemoryBase Model=Gemini-2.5-flash, Setting=Realistic (without any pre-filtering)2026.05 | 20.5 | 46.9 | 23.1 | 2 |