Policy Question Answering on PolicyBench US
66.43AccuracyDeepseek-R1
Evaluation results
| Method | Links | |
|---|---|---|
| Deepseek-R12026.04 | 66.43 | |
| o4-mini2026.04 | 65.54 | |
| Claude-3.5-Sonnet2026.04 | 65.39 | |
| Claude-3.7-sonnet2026.04 | 65.06 | |
| Gemini-2.5-Flash2026.04 | 64.03 | |
| GPT-4o2026.04 | 61.41 | |
| Gemini-2.0-Flash2026.04 | 60.84 | |
| Gemma 3-27B2026.04 | 60.15 | |
| LLaMA-42026.04 | 60.04 | |
| Deepseek-V32026.04 | 59.38 | |
| QwQ-32B2026.04 | 58 |