MMOOC - Image-Question Mismatch (Open-Ended VQA): leaderboard
Metric: Shifted in-context score (%; mean of accuracy on the answerable Image-Question Mismatch Open-Ended VQA questions and answer rationality, the mean 0-1 judge score of the answer's correctness and evidence consistency in steps of 0.25; judge scores averaged over GPT-5.6, Claude Opus 5 and DeepSeek-V4-Pro). Source: arxiv.org. Saturation forecast: Around December 2026. 18 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3.5 27B | 89 |
| 2 | Qwen 3.5 122B A10B | 86.75 |
| 3 | O3 | 85.25 |
| 4 | O1 | 81 |
| 5 | Gemma 4 31B | 80.5 |
| 6 | Ministral 3 8B | 80.5 |
| 7 | GPT-4o | 80.25 |
| 8 | Llama 4 Maverick | 80 |
| 9 | Gemma 4 26B | 78 |
| 10 | Ministral 3 14B | 73.5 |
| 11 | InternVL3-8B | 70.75 |
| 12 | Gemini 3.1 Pro (Preview) | 52 |
| 13 | Claude Opus 4.6 | 35.75 |
Interactive version: theaggregate.ai/benchmark?slug=mmooc-image-question-mismatch-open-ended-vqa · How It Works · Data refreshed daily, snapshot 2026-09-29.