Reasoning on HMMT
85Pass@1Vanilla
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| VanillaBackbone=Qwen3-235B2026.02 | 85 | 29,100 | |
| Free()LMBackbone=Qwen3-235B2026.02 | 84.58 | 21,700 | |
| GeminiBackbone=Qwen3-235B2026.02 | 84.17 | 25,600 | |
| No TrainBackbone=Qwen3-235B2026.02 | 80 | 24,400 | |
| Free()LMBackbone=Qwen3-30B2026.02 | 70.42 | 19,100 | |
| GeminiBackbone=Qwen3-30B2026.02 | 65.42 | 19,000 | |
| No TrainBackbone=Qwen3-30B2026.02 | 63.33 | 20,400 | |
| VanillaBackbone=Qwen3-30B2026.02 | 60.83 | 21,000 | |
| Free()LMBackbone=Qwen3-8B2026.02 | 49.58 | 16,400 | |
| CompassMax-V3-Thinking2025.12 | 46.7 | — | |
| H2OBackbone=Qwen3-8B2026.02 | 46.67 | 21,500 | |
| GeminiBackbone=Qwen3-8B2026.02 | 44.17 | 18,400 | |
| No TrainBackbone=Qwen3-8B2026.02 | 42.5 | 19,900 | |
| DeepSeek-R12025.12 | 41.67 | — | |
| VanillaBackbone=Qwen3-8B2026.02 | 38.75 | 19,400 | |
| TCBackbone=Qwen3-8B2026.02 | 26.67 | 20,500 | |
| CompassMax-V32025.12 | 11.52 | — |