ODCV-Bench - Mandated: leaderboard
Same 40 scenarios with the outcome explicitly demanded in the instruction; the number is the share of runs judged metric gaming or worse.
Metric: Misalignment rate (%, lower is better). Source: github.com. 21 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.7 | 0 | #79 |
| 2 | Claude Opus 4.5 | 0 | #156 |
| 3 | GPT-5.1 Instant | 5.13 | #440 |
| 4 | Llama 4 Maverick | 16.67 | #1099 |
| 5 | GPT-5.5 | 17.5 | #49 |
| 6 | GLM-5.1 | 22.5 | #179 |
| 7 | GLM-4.6 | 25 | #486 |
| 8 | Qwen 3.6 Max Preview | 30 | #159 |
| 9 | GPT-OSS-20B | 35 | #1147 |
| 10 | Kimi K2.6 | 37.5 | #178 |
| 11 | Kimi K2 0905 | 40 | #558 |
| 12 | Grok 4.1 Fast | 40 | #430 |
| 13 | MiniMax-M2.7 | 41.03 | #518 |
| 14 | Gemini 3.1 Pro (Preview) | 42.5 | #94 |
| 15 | Qwen 3 Max | 42.5 | #403 |
Interactive version: theaggregate.ai/benchmark?slug=odcv-bench-mandated · How It Works · Data refreshed daily, snapshot 2026-11-13.