FaithformBench - OlympiadBench - Unfaithfulness Lower Bound: leaderboard
Metric: Unfaithfulness lower bound (%; half FNR + half FPR + AF failure rate). Source: arxiv.org. 7 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.7 | 13 |
| 2 | Gemini 3.1 Pro (Preview) | 16 |
| 3 | GPT-5.2 | 26 |
Interactive version: theaggregate.ai/benchmark?slug=faithformbench-olympiadbench-unfaithfulness-lower-bound · How It Works · Data refreshed daily, snapshot 2026-09-19.