ERR-EVAL - Unclear Wording: leaderboard
The "Unclear Wording" track of ERR-EVAL, which scores whether a model notices a defective prompt and says what it cannot answer instead of inventing it. This track feeds ambiguous phrasing with more than one reading.
Metric: Track score. Source: err-eval.gustycube.com. Saturation forecast: Around April 2027. 60 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.2 Pro | 7.4 | #70 |
| 2 | GPT-5.4 | 7 | #83 |
| 3 | GPT-5.2 | 7 | #115 |
| 4 | Claude Opus 4.5 | 6.8 | #87 |
| 5 | Claude Sonnet 4.6 | 6.6 | #94 |
| 6 | Claude Opus 4.6 | 6.4 | #67 |
| 7 | Grok 4 | 6.4 | #189 |
| 8 | GPT-5.3 Codex | 6.2 | #72 |
| 9 | Claude Sonnet 4.5 | 6 | #148 |
| 10 | GPT-5.4 Mini | 6 | #228 |
| 11 | O3 | 6 | #132 |
| 12 | GPT-5.3 | 6 | #104 |
| 13 | Claude Opus 4 | 5.8 | #170 |
| 14 | Grok 4.20 Multi-Agent | 5.8 | #134 |
| 15 | Kimi K2.5 | 5.6 | #152 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=err-eval-unclear-wording · How It Works · Data refreshed daily, snapshot 2026-11-05.