Coding on LiveCodeBench (p@1, p@16)
69.11Pass@1LED
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| LEDModel=Qwen3-30B-A3B-Thinking2026.02 | 69.11 | 82.8 | |
| DoLaModel=Qwen3-30B-A3B-Thinking2026.02 | 68.35 | 79.21 | |
| ST-GModel=Qwen3-30B-A3B-Thinking2026.02 | 68.03 | 82.08 | |
| CoTModel=Qwen3-30B-A3B-Thinking2026.02 | 67.79 | 81.36 | |
| STModel=Qwen3-30B-A3B-Thinking2026.02 | 66.33 | 76.34 | |
| SwiRBackbone=Qwen3-8B2025.10 | 63.44 | — | |
| LEDModel=Qwen3-4B-Thinking2026.02 | 63.1 | 77.06 | |
| ST-GModel=Qwen3-4B-Thinking2026.02 | 62.57 | 77.02 | |
| CoTBackbone=Qwen3-8B2025.10 | 62.01 | — | |
| CoTModel=Qwen3-4B-Thinking2026.02 | 61.76 | 75.27 | |
| DoLaModel=Qwen3-4B-Thinking2026.02 | 61.25 | 75.27 | |
| STModel=Qwen3-4B-Thinking2026.02 | 59.41 | 71.33 | |
| Soft ThinkingBackbone=Qwen3-8B2025.10 | 56.99 | — | |
| LEDModel=MiMo-7B-RL2026.02 | 56.09 | 72.4 | |
| CoTModel=MiMo-7B-RL2026.02 | 56.01 | 71.68 | |
| DoLaModel=MiMo-7B-RL2026.02 | 55.58 | 73.12 | |
| ST-GModel=MiMo-7B-RL2026.02 | 54.57 | 72.4 | |
| STModel=MiMo-7B-RL2026.02 | 54.39 | 72.76 | |
| CoT (Greedy)Backbone=Qwen3-8B, decoding=Greedy2025.10 | 50.18 | — |