Mathematical Reasoning on OlympiadBench (EM)
57.9EMSPIRAL
Evaluation results
| Method | Links | |
|---|---|---|
| SPIRALFamily=DeepSeek-Distill-Qwen-7B Family, Game=Multi-Game2025.06 | 57.9 | |
| DeepSeek-Distill-Qwen-7BFamily=DeepSeek-Distill-Qwen-7B Family2025.06 | 56.9 | |
| SPIRALFamily=Qwen3-4B-Base Family, Game=Multi-Game2025.06 | 41.8 | |
| SPIRALFamily=Octothinker-8B-Base Family, Game=Multi-Game2025.06 | 33.7 | |
| Qwen3-4B-BaseFamily=Qwen3-4B-Base Family2025.06 | 33.3 | |
| Octothinker-8B-BaseFamily=Octothinker-8B-Base Family2025.06 | 26.6 | |
| GHS-TDABackbone=DeepSeek-V32026.02 | 14.7 | |
| GHS-TDABackbone=GPT-4o-mini2026.02 | 14.5 | |
| GHS-TDABackbone=Qwen-Turbo2026.02 | 14.4 | |
| GoTBackbone=DeepSeek-V3, Reasoning Paradigm=graph-based2026.02 | 13.7 | |
| GoTBackbone=GPT-4o-mini, Reasoning Paradigm=graph-based2026.02 | 13.1 | |
| AoTBackbone=DeepSeek-V3, Reasoning Paradigm=atomic reasoning2026.02 | 13.1 | |
| GoTBackbone=Qwen-Turbo, Reasoning Paradigm=graph-based2026.02 | 13 | |
| AoTBackbone=Qwen-Turbo, Reasoning Paradigm=atomic reasoning2026.02 | 12.6 | |
| FoTBackbone=DeepSeek-V3, Reasoning Paradigm=forest-based, n=82026.02 | 12.6 | |
| FoTBackbone=GPT-4o-mini, Reasoning Paradigm=forest-based, n=82026.02 | 12.5 | |
| AFlowBackbone=DeepSeek-V3, Reasoning Paradigm=search-based2026.02 | 12.5 | |
| AFlowBackbone=GPT-4o-mini, Reasoning Paradigm=search-based2026.02 | 12.4 | |
| FoTBackbone=Qwen-Turbo, Reasoning Paradigm=forest-based, n=82026.02 | 12.3 | |
| AoTBackbone=GPT-4o-mini, Reasoning Paradigm=atomic reasoning2026.02 | 12.1 | |
| AFlowBackbone=Qwen-Turbo, Reasoning Paradigm=search-based2026.02 | 12.1 | |
| ToTBackbone=DeepSeek-V3, Reasoning Paradigm=tree-based2026.02 | 11.6 | |
| ToTBackbone=GPT-4o-mini, Reasoning Paradigm=tree-based2026.02 | 11.4 | |
| ToTBackbone=Qwen-Turbo, Reasoning Paradigm=tree-based2026.02 | 11.3 | |
| CoT-SCBackbone=DeepSeek-V3, Reasoning Paradigm=chain-based, n=52026.02 | 10.4 | |
| CoT-SCBackbone=GPT-4o-mini, Reasoning Paradigm=chain-based, n=52026.02 | 10.2 | |
| CoT-SCBackbone=Qwen-Turbo, Reasoning Paradigm=chain-based, n=52026.02 | 9.9 | |
| CoTBackbone=DeepSeek-V3, Reasoning Paradigm=chain-based2026.02 | 9.5 | |
| Self-RefineBackbone=DeepSeek-V3, Reasoning Paradigm=chain-based2026.02 | 9.5 | |
| Self-RefineBackbone=GPT-4o-mini, Reasoning Paradigm=chain-based2026.02 | 9.4 | |
| Self-RefineBackbone=Qwen-Turbo, Reasoning Paradigm=chain-based2026.02 | 9.4 | |
| CoTBackbone=GPT-4o-mini, Reasoning Paradigm=chain-based2026.02 | 9.3 | |
| CoTBackbone=Qwen-Turbo, Reasoning Paradigm=chain-based2026.02 | 8.9 | |
| Analogical PromptingBackbone=DeepSeek-V3, Reasoning Paradigm=chain-based2026.02 | 6.7 | |
| Analogical PromptingBackbone=GPT-4o-mini, Reasoning Paradigm=chain-based2026.02 | 6.5 | |
| Analogical PromptingBackbone=Qwen-Turbo, Reasoning Paradigm=chain-based2026.02 | 6.2 |