List your company

Mathematical Reasoning on AIME 24 (Accuracy, Think Tokens)

18,923Thinking TokensRaw Model
9,812.612,177.814,54316,908.2Jun 16, 2026

Evaluation results

MethodLinks
Raw ModelBackbone=Qwen3-4B-Thinking-25072026.0618,92381.87
DEERBackbone=Qwen3-4B-Thinking-25072026.0618,62080.74
Raw ModelBackbone=Qwen3-30B-A3B-Thinking-25072026.0615,29988.12
DEERBackbone=Qwen3-30B-A3B-Thinking-25072026.0614,40587.78
RCPDBackbone=Qwen3-4B-Thinking-25072026.0614,38479.37
Raw ModelBackbone=Qwen3-8B2026.0613,98875.42
RCPDBackbone=Qwen3-30B-A3B-Thinking-25072026.0613,83187.08
DREBackbone=Qwen3-4B-Thinking-25072026.0613,09281.87
RCPDBackbone=Qwen3-8B2026.0612,95274.79
DEERBackbone=Qwen3-8B2026.0612,21972.08
DREBackbone=Qwen3-30B-A3B-Thinking-25072026.0610,40788.54
DREBackbone=Qwen3-8B2026.0610,16375.83