General Reasoning on Average (AIME24, AIME25, AIME26, GPQA-D)
10,230Think TokensDRE
Evaluation results
| Method | Links | ||
|---|---|---|---|
| DREBackbone=Qwen3-30B-A3B-Thinking-25072026.06 | 10,230 | 83.51 | |
| DREBackbone=Qwen3-8B2026.06 | 10,413 | 67.59 | |
| DREBackbone=Qwen3-4B-Thinking-25072026.06 | 12,024 | 76.86 | |
| DEERBackbone=Qwen3-8B2026.06 | 12,344 | 64.12 | |
| RCPDBackbone=Qwen3-8B2026.06 | 12,618 | 65.48 | |
| RCPDBackbone=Qwen3-30B-A3B-Thinking-25072026.06 | 12,706 | 82.85 | |
| RCPDBackbone=Qwen3-4B-Thinking-25072026.06 | 13,653 | 74.95 | |
| Raw ModelBackbone=Qwen3-8B2026.06 | 13,967 | 67.55 | |
| DEERBackbone=Qwen3-30B-A3B-Thinking-25072026.06 | 14,183 | 83.4 | |
| Raw ModelBackbone=Qwen3-30B-A3B-Thinking-25072026.06 | 14,325 | 84.79 | |
| DEERBackbone=Qwen3-4B-Thinking-25072026.06 | 16,601 | 75.85 | |
| Raw ModelBackbone=Qwen3-4B-Thinking-25072026.06 | 17,328 | 76.21 |