IMAVB (Fixed Option Order): leaderboard
The IMAVB (Nguyen Quang et al., 2026) conflict-detection questions with the options in canonical order, so the two reject options (incorrect visual detail, incorrect audio detail) are always E and F. Balanced accuracy in percent over the four standard and misleading splits of 500 items each; higher is better.
Metric: Balanced accuracy (%; mean of the four split accuracies, 2,000 items, six-option MCQ with the reject options E and F always last). Source: arxiv.org. Saturation forecast: Around February 2027. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 61.6 | #91 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=imavb-fixed-option-order · How It Works · Data refreshed daily, snapshot 2026-11-12.