List your company

Graduate-level Science QA on GPQA D

5,318Tokens Used for ThinkingDRE
5,144.326,316.667,4898,661.34Jun 16, 2026

Evaluation results

MethodLinks
DREBase Model=Qwen3-30B-A3B-Thinking-25072026.065,31875.51
DREBase Model=Qwen3-4B-Thinking-25072026.065,68965.67
GRPOBase Model=Qwen3-30B-A3B-Thinking-25072026.066,23674.68
GRPO + LPBase Model=Qwen3-30B-A3B-Thinking-25072026.066,43575
DREBase Model=Qwen3-8B2026.066,65860.98
Raw ModelBase Model=Qwen3-30B-A3B-Thinking-25072026.066,92276.45
S-GRPOBase Model=Qwen3-30B-A3B-Thinking-25072026.067,00074.68
GRPOBase Model=Qwen3-4B-Thinking-25072026.067,82665.26
GRPO + LPBase Model=Qwen3-4B-Thinking-25072026.067,90666.6
Raw ModelBase Model=Qwen3-4B-Thinking-25072026.068,09165.47
GRPO + LPBase Model=Qwen3-8B2026.068,15960.1
S-GRPOBase Model=Qwen3-4B-Thinking-25072026.068,43565.59
GRPOBase Model=Qwen3-8B2026.068,47460.61
Raw ModelBase Model=Qwen3-8B2026.068,68359.34
S-GRPOBase Model=Qwen3-8B2026.069,66059.79