Mathematical Reasoning on GSM8K (Acc, Tokens)
97AccuracyCOT
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| COTModel=QwQ-32B2026.04 | 97 | 1,561 | — | |
| COTModel=QwQ-32B2026.04 | 97 | 1,561 | — | |
| EquivPrunerBackbone=Qwen2.5-14B-Instruct2025.05 | 97 | 27,402 | 61.71 | |
| BaselineBackbone=QwQ-32B2026.06 | 96.8 | 1,505 | — | |
| DyConBackbone=QwQ-32B2026.06 | 96.8 | 995 | — | |
| CyclicReflexBase Model=QwQ-32B2026.05 | 96.7 | 1,581 | — | |
| PathCalBase Model=QwQ-32B2026.05 | 96.7 | 1,430 | — | |
| NoWaitBackbone=QwQ-32B2026.06 | 96.7 | 983 | — | |
| QwQ-32B MixReasoningMode=MixReasoning, Base Model=QwQ-32B2025.10 | 96.64 | 880 | — | |
| SATModel=QwQ-32B2026.04 | 96.6 | 969 | — | |
| SATModel=QwQ-32B2026.04 | 96.6 | 969 | — | |
| Qwen3-14B MixReasoningMode=MixReasoning, Base Model=Qwen3-14B2025.10 | 96.57 | 1,137 | — | |
| SATModel=Qwen3-32B2026.04 | 96.5 | 680 | — | |
| SATModel=Qwen3-32B2026.04 | 96.5 | 680 | — | |
| CoDBackbone=QwQ-32B2026.06 | 96.5 | 617 | — | |
| NothinkingBackbone=QwQ-32B2026.06 | 96.5 | 1,331 | — | |
| FlashThinkBackbone=QwQ-32B2026.06 | 96.5 | 910 | — | |
| SATModel=Qwen3-14B2026.04 | 96.4 | 767 | — | |
| SATModel=Qwen3-14B2026.04 | 96.4 | 767 | — | |
| TIPBase Model=QwQ-32B2026.05 | 96.4 | 1,494 | — | |
| S1Base Model=QwQ-32B2026.05 | 96.4 | 1,830 | — | |
| QwQ-32BMode=Original, Base Model=QwQ-32B2025.10 | 96.37 | 1,745 | — | |
| DEERModel=QwQ-32B2026.04 | 96.3 | 977 | — | |
| DEERModel=QwQ-32B2026.04 | 96.3 | 977 | — | |
| DEERBackbone=QwQ-32B2026.06 | 96.3 | 977 | — | |
| SEALBackbone=QwQ-32B2026.06 | 96.3 | 1,231 | — | |
| BaselineBackbone=Qwen3-14B2026.06 | 96.3 | 1,693 | — | |
| DyConBackbone=Qwen3-14B2026.06 | 96.3 | 1,166 | — | |
| Qwen3-14BMode=Original, Base Model=Qwen3-14B2025.10 | 96.25 | 1,950 | — | |
| DEERModel=Qwen3-32B2026.04 | 96.2 | 769 | — | |
| DEERModel=Qwen3-32B2026.04 | 96.2 | 769 | — | |
| CoDBackbone=Qwen3-14B2026.06 | 96.2 | 670 | — | |
| OriginalBase Model=QwQ-32B2026.05 | 96.1 | 1,593 | — | |
| MERABackbone=DeepSeek-R1-Distill-Qwen-14B, Zero-shot Chain-of-Thought (CoT)=true2025.08 | 96.1 | 835 | — | |
| COTModel=Qwen3-32B2026.04 | 96 | 1,688 | — | |
| COTModel=Qwen3-32B2026.04 | 96 | 1,688 | — | |
| Vanilla MCTSBackbone=Qwen2.5-14B-Instruct2025.05 | 96 | 44,404 | 100 | |
| BaselineBackbone=Qwen3-4B-Thinking-25072026.06 | 95.9 | 1,494 | — | |
| Qwen3-8B MixReasoningMode=MixReasoning, Base Model=Qwen3-8B2025.10 | 95.88 | 1,389 | — | |
| COTModel=Qwen3-8B2026.04 | 95.8 | 2,152 | — | |
| COTModel=Qwen3-8B2026.04 | 95.8 | 2,152 | — | |
| NoWaitBackbone=Qwen3-14B2026.06 | 95.8 | 1,125 | — | |
| COTModel=Qwen3-14B2026.04 | 95.7 | 1,660 | — | |
| COTModel=Qwen3-14B2026.04 | 95.7 | 1,660 | — | |
| CoDBackbone=Qwen3-4B-Thinking-25072026.06 | 95.7 | 952 | — | |
| DEERBackbone=Qwen3-4B-Thinking-25072026.06 | 95.7 | 1,037 | — | |
| DyConBackbone=Qwen3-4B-Thinking-25072026.06 | 95.7 | 1,098 | — | |
| Dynasor-CoTBackbone=Qwen3-14B2026.06 | 95.6 | 1,483 | — | |
| QwQ-32B Concise ModeMode=Concise Mode, alpha=32, Base Model=QwQ-32B2025.10 | 95.58 | 664 | — | |
| DEERModel=Qwen3-8B2026.04 | 95.5 | 981 | — | |
| SATModel=Qwen3-8B2026.04 | 95.5 | 879 | — | |
| DEERModel=Qwen3-8B2026.04 | 95.5 | 981 | — | |
| SATModel=Qwen3-8B2026.04 | 95.5 | 879 | — | |
| Qwen3-8BMode=Original, Base Model=Qwen3-8B2025.10 | 95.45 | 2,403 | — | |
| DEERModel=Qwen3-14B2026.04 | 95.3 | 840 | — | |
| DEERModel=Qwen3-14B2026.04 | 95.3 | 840 | — | |
| DEERBackbone=Qwen3-14B2026.06 | 95.3 | 840 | — | |
| SATModel=DS-Qwen-14B2026.04 | 95.2 | 621 | — | |
| SATModel=DS-Qwen-14B2026.04 | 95.2 | 621 | — | |
| QwQ-32B Concise ModeMode=Concise Mode, alpha=64, Base Model=QwQ-32B2025.10 | 95.14 | 391 | — | |
| COTModel=Qwen3-4B2026.04 | 95.1 | 2,136 | — | |
| SATModel=Qwen3-4B2026.04 | 95.1 | 845 | — | |
| COTModel=Qwen3-4B2026.04 | 95.1 | 2,136 | — | |
| SATModel=Qwen3-4B2026.04 | 95.1 | 845 | — | |
| NothinkingBackbone=Qwen3-4B-Thinking-25072026.06 | 95 | 1,137 | — | |
| Qwen3-14B Concise ModeMode=Concise Mode, alpha=32, Base Model=Qwen3-14B2025.10 | 95 | 674 | — | |
| COTModel=DS-Qwen-14B2026.04 | 94.9 | 1,122 | — | |
| COTModel=DS-Qwen-14B2026.04 | 94.9 | 1,122 | — | |
| NothinkingBackbone=Qwen3-14B2026.06 | 94.9 | 260 | — | |
| ThinkpilotBackbone=Qwen3-14B2026.06 | 94.9 | 274 | — | |
| Qwen3-14B Concise ModeMode=Concise Mode, alpha=64, Base Model=Qwen3-14B2025.10 | 94.87 | 372 | — | |
| NoWaitBackbone=Qwen3-4B-Thinking-25072026.06 | 94.8 | 1,070 | — | |
| DASTBackbone=DeepSeek-R1-Distill-Qwen-14B, Zero-shot Chain-of-Thought (CoT)=true2025.08 | 94.8 | 1,778 | — | |
| ThinkpilotBackbone=Qwen3-4B-Thinking-25072026.06 | 94.7 | 878 | — | |
| FCS+Ref.Backbone=DeepSeek-R1-Distill-Qwen-14B, Zero-shot Chain-of-Thought (CoT)=true2025.08 | 94.7 | 1,141 | — | |
| DEERModel=Qwen3-4B2026.04 | 94.5 | 1,250 | — | |
| DEERModel=Qwen3-4B2026.04 | 94.5 | 1,250 | — | |
| ThinkSwitcherModel=DS-Qwen-14B2026.04 | 94.3 | 1,042 | — | |
| ThinkSwitcherModel=DS-Qwen-14B2026.04 | 94.3 | 1,042 | — | |
| S1Base Model=DeepSeek-R1-Distill-Qwen-14B2026.05 | 93.9 | 607 | — | |
| TrimRBackbone=QwQ-32B2026.06 | 93.7 | 1,319 | — | |
| MERABackbone=DeepSeek-R1-Distill-Qwen-7B, Zero-shot Chain-of-Thought (CoT)=true2025.08 | 93.7 | 822 | — | |
| Qwen3-8B Concise ModeMode=Concise Mode, alpha=32, Base Model=Qwen3-8B2025.10 | 93.65 | 817 | — | |
| PathCalBase Model=DeepSeek-R1-Distill-Qwen-14B2026.05 | 93.6 | 479 | — | |
| CyclicReflexBase Model=DeepSeek-R1-Distill-Qwen-14B2026.05 | 93.5 | 537 | — | |
| DEERBackbone=DeepSeek-R1-Distill-Qwen-14B, Zero-shot Chain-of-Thought (CoT)=true2025.08 | 93.5 | 975 | — | |
| DEERModel=DS-Qwen-14B2026.04 | 93.3 | 1,040 | — | |
| DEERModel=DS-Qwen-14B2026.04 | 93.3 | 1,040 | — | |
| OriginalBase Model=DeepSeek-R1-Distill-Qwen-14B2026.05 | 93.3 | 480 | — | |
| TIPBase Model=DeepSeek-R1-Distill-Qwen-14B2026.05 | 93 | 482 | — | |
| FCS+Ref.Backbone=DeepSeek-R1-Distill-Qwen-7B, Zero-shot Chain-of-Thought (CoT)=true2025.08 | 92.9 | 1,218 | — | |
| Qwen3-8B Concise ModeMode=Concise Mode, alpha=64, Base Model=Qwen3-8B2025.10 | 92.89 | 534 | — | |
| DASTBackbone=DeepSeek-R1-Distill-Qwen-7B, Zero-shot Chain-of-Thought (CoT)=true2025.08 | 92.7 | 1,558 | — | |
| No WaitBackbone=DeepSeek-R1-Distill-Qwen-14B, Zero-shot Chain-of-Thought (CoT)=true2025.08 | 92.6 | 1,152 | — | |
| ThinkSwitcherModel=DS-Qwen-7B2026.04 | 92.5 | 1,389 | — | |
| ThinkSwitcherModel=DS-Qwen-7B2026.04 | 92.5 | 1,389 | — | |
| O1-PrunerBackbone=DeepSeek-R1-Distill-Qwen-7B, Zero-shot Chain-of-Thought (CoT)=true2025.08 | 92.5 | 1,052 | — | |
| DeepSeek-R1-7B MixReasoningMode=MixReasoning, Base Model=DeepSeek-R1-7B2025.10 | 92.3 | 765 | — | |
| OriginalBackbone=DeepSeek-R1-Distill-Qwen-14B, Zero-shot Chain-of-Thought (CoT)=true2025.08 | 92.3 | 1,917 | — | |
| DeepSeek-R1-7BMode=Original, Base Model=DeepSeek-R1-7B2025.10 | 92.03 | 1,214 | — |