Web-based Agent Reasoning on WebWalkerQA Hard
0.6333Pass@3ExpSeek
Evaluation Results
| Method | Links | |
|---|---|---|
| ExpSeekBase Model=Qwen3-8B2026.01 | 0.6333 | |
| ExpSeekBase Model=Qwen3-32B2026.01 | 0.6333 | |
| No ExperienceBase Model=Qwen3-32B2026.01 | 0.5833 | |
| Training-Free GRPOBase Model=Qwen3-32B2026.01 | 0.5745 | |
| REASONINGBANK+Base Model=Qwen3-32B2026.01 | 0.5644 | |
| Training-Free GRPOBase Model=Qwen3-8B2026.01 | 0.5267 | |
| REASONINGBANK+Base Model=Qwen3-8B2026.01 | 0.5 | |
| No ExperienceBase Model=Qwen3-8B2026.01 | 0.4611 |