Agent Reasoning on xbench (test)
0.66Pass@3ExpSeek
Evaluation Results
| Method | Links | |
|---|---|---|
| ExpSeekBase Model=Qwen3-32B2026.01 | 0.66 | |
| ExpSeekBase Model=Qwen3-8B2026.01 | 0.62 | |
| REASONINGBANK+Base Model=Qwen3-32B2026.01 | 0.59 | |
| Training-Free GRPOBase Model=Qwen3-32B2026.01 | 0.536 | |
| No ExperienceBase Model=Qwen3-32B2026.01 | 0.53 | |
| REASONINGBANK+Base Model=Qwen3-8B2026.01 | 0.468 | |
| Training-Free GRPOBase Model=Qwen3-8B2026.01 | 0.45 | |
| No ExperienceBase Model=Qwen3-8B2026.01 | 0.43 |