Zebra Logic Puzzle Solving on Zebra Logic Solvable
94.5AccuracyDeepseek-V3.2-R
Evaluation Results
| Method | Links | |
|---|---|---|
| Deepseek-V3.2-RModel=Deepseek-V3.2-R2025.12 | 94.5 | |
| Gemini-3.0-ProModel=Gemini-3.0-Pro2025.12 | 65 | |
| GPT-5.1-LowModel=GPT-5.1-Low2025.12 | 48 | |
| Qwen3-1.7B-InstructModel=Qwen3-1.7B-Instruct2025.12 | 9 | |
| Qwen3-4B-Instruct + UnsolRL-FinalBase Model=Qwen3-4B-Instruct, Method=UnsolRL-Final2025.12 | 8.8 | |
| Qwen3-1.7B-Instruct + UnsolRL-FinalBase Model=Qwen3-1.7B-Instruct, Method=UnsolRL-Final2025.12 | 5.3 | |
| Qwen3-4B-InstructModel=Qwen3-4B-Instruct2025.12 | 2 |