Scientific Fact-Checking on BIONLI 300
68.7Balanced AccuracyAtomic+Search
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Atomic+Searchsearch=enabled2026.04 | 68.7 | 65.4 | 66.7 | |
| GPT-5 Mini + Searchsearch=enabled2026.04 | 66.9 | 62.7 | 65.8 | |
| RARR2026.04 | 66.4 | 62.3 | 65.3 | |
| GPT-o12026.04 | 65.8 | 60.2 | 64.9 | |
| GPT-5 Mini2026.04 | 62.9 | 58.7 | 61.8 | |
| Qwen 32B MAD2026.04 | 62.5 | 59.1 | 61.3 | |
| MiniCheck2026.04 | 61.35 | 59.8 | 60.7 |