List your company

Mathematical Reasoning on AIME 24 (Pass@1, #Tok)

86.7Pass@1 AccuracyDyCon
10.36430.1825069.818Jun 5, 2026

Evaluation results

MethodLinks
DyConBackbone=Qwen3-4B-Thinking-25072026.0686.718,867
BaselineBackbone=Qwen3-4B-Thinking-25072026.0683.321,493
CoDBackbone=Qwen3-4B-Thinking-25072026.0683.318,652
BaselineBackbone=Qwen3-14B2026.0676.712,746
DEERBackbone=Qwen3-14B2026.0676.77,619
NoWaitBackbone=Qwen3-14B2026.0676.710,181
DyConBackbone=Qwen3-14B2026.0676.710,536
NothinkingBackbone=Qwen3-4B-Thinking-25072026.0673.316,556
BaselineBackbone=QwQ-32B2026.0673.313,364
NoWaitBackbone=QwQ-32B2026.0673.39,405
DyConBackbone=QwQ-32B2026.0673.312,794
Dynasor-CoTBackbone=Qwen3-14B2026.0673.310,369
DEERBackbone=QwQ-32B2026.067010,087
DEERBackbone=Qwen3-4B-Thinking-25072026.0666.712,728
NothinkingBackbone=QwQ-32B2026.0666.711,507
CoDBackbone=QwQ-32B2026.0663.311,029
SEALBackbone=QwQ-32B2026.0663.312,064
CoDBackbone=Qwen3-14B2026.0663.311,426
FlashThinkBackbone=QwQ-32B2026.066010,034
TrimRBackbone=QwQ-32B2026.0656.78,345
CoDBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0653.311,419
Manifold SteeringBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0653.38,457
DyConBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0653.310,906
NoWaitBackbone=Qwen3-4B-Thinking-25072026.0653.312,393
BaselineBackbone=DeepSeek-R1-Distill-Qwen-7B2026.065013,008
Controlling Thinking SpeedBackbone=DeepSeek-R1-Distill-Qwen-7B2026.065012,588
DEERBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0649.29,839
SEALBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0643.311,092
ThinkpilotBackbone=Qwen3-4B-Thinking-25072026.0643.37,913
NoWaitBackbone=DeepSeek-R1-Distill-Qwen-7B2026.06407,281
NothinkingBackbone=Qwen3-14B2026.06305,123
ThinkpilotBackbone=Qwen3-14B2026.0626.77,841
NothinkingBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0616.74,222
ThinkpilotBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0613.31,229