Tool Attribution Correctness on Wiki Multihop QA
91.9PrecisionALCE Vanilla GPT5-Nano
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| ALCE Vanilla GPT5-NanoAttribution Category=True2026.02 | 91.9 | 78.2 | 84.5 | |
| AIF GPT5-NanoAttribution Category=True2026.02 | 80.1 | 76.8 | 78.4 | |
| AIF GPT5-NanoAttribution Category=False2026.02 | 57.4 | 56.4 | 56.9 | |
| ALCE Vanilla GPT5-NanoAttribution Category=False2026.02 | 53.5 | 53.8 | 53.7 |