Tool Attribution Correctness on MSMARCO
39.3PrecisionAIF GPT5-Nano
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| AIF GPT5-NanoAttribution Category=True2026.02 | 39.3 | 87.2 | 54.2 | |
| ALCE Vanilla GPT5-NanoAttribution Category=True2026.02 | 35.1 | 81.6 | 49.1 | |
| AIF GPT5-NanoAttribution Category=False2026.02 | 8.4 | 86.6 | 15.3 | |
| ALCE Vanilla GPT5-NanoAttribution Category=False2026.02 | 7.3 | 84.9 | 13.5 |