Modality Arbitration - PhD-icc - Vanilla Image Accuracy: leaderboard
PhD-icc,1000examples,vanillaQwen3VL8BInstructBF16,greedy128tokens,image-targetaccuracy. Text-preference and steering settings are separate.
Metric: Score (%). Source: arxiv.org. 1 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen3-VL-8B-Instruct | 64.4 |
Interactive version: theaggregate.ai/benchmark?slug=modality-arbitration-phd-icc-vanilla-image-accuracy · How It Works · Data refreshed daily, snapshot 2026-11-12.