Full-Text Screening on Review 1
18False PositivesClaude sonnet
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Claude sonnet2025.12 | 18 | — | 31.41 | 97.11 | 81.7 | 76.92 | 0.66 | 0.16 | 0.05 | |
| gpt-52025.12 | 30 | — | 57.07 | 94.94 | 85.71 | 78.42 | 0.78 | 0.13 | 0.08 | |
| grok-4-fast2025.12 | 42 | — | 58.64 | 93.26 | 85.14 | 72.73 | 0.76 | 0.22 | 0.18 | |
| gpt-5-mini2025.12 | 82 | — | 73.3 | 86.73 | 83.56 | 63.06 | 0.75 | 0.22 | 0.19 | |
| gpt-4o-mini2025.12 | 88 | — | 55.5 | 85.8 | 78.67 | 54.64 | 0.51 | 0.37 | 0.51 | |
| Gemini flash2025.12 | 95 | — | 73.3 | 84.45 | 81.8 | 59.57 | 0.69 | 0.35 | 0.28 | |
| gpt-5-nano2025.12 | 168 | — | 80.63 | 72.95 | 74.75 | 47.83 | 0.73 | 0.28 | 0.25 | |
| Claude haiku2025.12 | 169 | — | 68.06 | 72.02 | 71.07 | 43.48 | 0.61 | 0.48 | 0.45 |