List your company

Scientific Reasoning on ChemBench (test)

66Initial ScoreQWEN3-MAX
49.8854.06558.2562.435May 11, 2026

Evaluation results

MethodLinks
QWEN3-MAXModel Scale=Max2026.056611.50.567
GRPOModel Scale=9B2026.0563-2.502.560.5
DPOModel Scale=9B2026.0561.503.53.561.5
ReCritModel Scale=9B2026.0561.5812.54.569.5
CRITIQUEGRPOModel Scale=9B2026.0560.524262.5
BASE MODELModel Scale=9B2026.0559.50.532.560
SFTModel Scale=9B2026.0559.514360.5
GRPOModel Scale=4B2026.0556-1.512.554.5
SFTModel Scale=4B2026.0555.5-0.51.5255
DPOModel Scale=4B2026.05550.52.5255.5
BASE MODELModel Scale=4B2026.0553.5-0.511.553
CRITIQUEGRPOModel Scale=4B2026.0552.51.542.554
ReCritModel Scale=4B2026.0550.510.516.5661