Mathematical Reasoning on AIME 24 (Accuracy, Think Tokens)
18,923Thinking TokensRaw Model
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Raw ModelBackbone=Qwen3-4B-Thinking-25072026.06 | 18,923 | 81.87 | |
| DEERBackbone=Qwen3-4B-Thinking-25072026.06 | 18,620 | 80.74 | |
| Raw ModelBackbone=Qwen3-30B-A3B-Thinking-25072026.06 | 15,299 | 88.12 | |
| DEERBackbone=Qwen3-30B-A3B-Thinking-25072026.06 | 14,405 | 87.78 | |
| RCPDBackbone=Qwen3-4B-Thinking-25072026.06 | 14,384 | 79.37 | |
| Raw ModelBackbone=Qwen3-8B2026.06 | 13,988 | 75.42 | |
| RCPDBackbone=Qwen3-30B-A3B-Thinking-25072026.06 | 13,831 | 87.08 | |
| DREBackbone=Qwen3-4B-Thinking-25072026.06 | 13,092 | 81.87 | |
| RCPDBackbone=Qwen3-8B2026.06 | 12,952 | 74.79 | |
| DEERBackbone=Qwen3-8B2026.06 | 12,219 | 72.08 | |
| DREBackbone=Qwen3-30B-A3B-Thinking-25072026.06 | 10,407 | 88.54 | |
| DREBackbone=Qwen3-8B2026.06 | 10,163 | 75.83 |