List your company

Agent Reasoning on xbench (test)

0.66Pass@3ExpSeek
0.42080.48290.5450.6071Jan 13, 2026

Evaluation results

MethodLinks
ExpSeekBase Model=Qwen3-32B2026.010.66
ExpSeekBase Model=Qwen3-8B2026.010.62
REASONINGBANK+Base Model=Qwen3-32B2026.010.59
Training-Free GRPOBase Model=Qwen3-32B2026.010.536
No ExperienceBase Model=Qwen3-32B2026.010.53
REASONINGBANK+Base Model=Qwen3-8B2026.010.468
Training-Free GRPOBase Model=Qwen3-8B2026.010.45
No ExperienceBase Model=Qwen3-8B2026.010.43