VEX-Bench - Justification Accuracy: leaderboard
Metric: Accuracy (%). Source: arxiv.org. 9 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.5 | 78.2 |
| 2 | Claude Opus 4.6 | 76.4 |
| 3 | DeepSeek V4 Pro | 76 |
| 4 | GLM-5.1 | 74.7 |
| 5 | Kimi K2.6 | 72.9 |
| 6 | Claude Sonnet 4.6 | 67.1 |
| 7 | GPT-5.4 Mini | 66.7 |
| 8 | MiniMax-M2.7 | 59.1 |
| 9 | DeepSeek V4 Flash | 58.7 |
Interactive version: theaggregate.ai/benchmark?slug=vex-bench-justification-accuracy · How It Works · Data refreshed daily, snapshot 2026-09-20.