Scientific Information Verification on Sci-PRMBench Search Information Verification
74.58AccuracySci-PRM
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Sci-PRM2026.06 | 74.58 | 75 | 53.85 | 62.69 | 88.2 | |
| Qwen3-VL-8B2026.06 | 62.08 | 52.58 | 43.22 | 47.44 | 74.44 | |
| GPT-5-Mini2026.06 | 59.32 | 47.73 | 17.8 | 25.93 | 87.01 | |
| Qwen3-VL-32B2026.06 | 57 | 43.42 | 27.73 | 33.85 | 76.24 | |
| Gemini-3-Flash2026.06 | 56.67 | 77.53 | 38.55 | 51.49 | 83.47 | |
| Doubao-Search2026.06 | 43.33 | 37.44 | 63.87 | 47.2 | 29.83 | |
| GLM-4.6V2026.06 | 42.14 | 37.78 | 72.03 | 49.56 | 22.65 | |
| Llama-3.2-11B-V2026.06 | 30.63 | 32.64 | 68.42 | 44.19 | 5.29 |