List your company

Scientific Reasoning on GPQA-D (Acc, Think Tok.)

5,318Thinking TokensDRE
5,183.46,091.957,000.57,909.05Jun 16, 2026

Evaluation results

MethodLinks
DREBackbone=Qwen3-30B-A3B-Thinking-25072026.065,31875.51
DREBackbone=Qwen3-4B-Thinking-25072026.065,68965.67
RCPDBackbone=Qwen3-30B-A3B-Thinking-25072026.066,02174.94
DEERBackbone=Qwen3-30B-A3B-Thinking-25072026.066,52875.47
DREBackbone=Qwen3-8B2026.066,65860.98
Raw ModelBackbone=Qwen3-30B-A3B-Thinking-25072026.066,92276.45
RCPDBackbone=Qwen3-4B-Thinking-25072026.067,26566.04
RCPDBackbone=Qwen3-8B2026.067,33059.85
DEERBackbone=Qwen3-4B-Thinking-25072026.067,91466.72
Raw ModelBackbone=Qwen3-4B-Thinking-25072026.068,09165.47
DEERBackbone=Qwen3-8B2026.068,13860.98
Raw ModelBackbone=Qwen3-8B2026.068,68359.34