MEGA-Bench Task - Red Teaming Visual Misleading: leaderboard
The "Red Teaming Visual Misleading" subtask of MEGA-Bench Task. A single task from MEGA-Bench, a large-scale multimodal evaluation suite that scores vision-language models across 500+ real-world tasks spanning perception, reasoning, OCR, charts, documents, and multi-image understanding.
Metric: Task Score (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 44 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 2 VL 7B | 90.5 | #750 |
| 2 | Qwen 2 VL 72B | 90.5 | #460 |
| 3 | Gemma 3 27B (IT) | 90 | #523 |
| 4 | Gemini 2.5 Pro (Preview 03-25) | 90 | #123 |
| 5 | MiniCPM-V-2.6 | 89.5 | #836 |
| 6 | Pixtral-12B | 89.5 | #808 |
| 7 | Claude 3.5 Sonnet (20241022) | 88.4 | #291 |
| 8 | Gemini 1.5 Pro (002) | 88.4 | #349 |
| 9 | Gemini 1.5 Flash (002) | 87.9 | #583 |
| 10 | Claude 3.5 Sonnet (20240620) | 87.9 | #299 |
| 11 | InternVL3-38B | 87.9 | #405 |
| 12 | Gemini 2.0 Flash (Preview) | 87.9 | #370 |
| 13 | InternVL2-8B | 86.8 | #817 |
| 14 | Qwen 2 VL 2B | 86.3 | #1220 |
| 15 | Gemma 3 4B (IT) | 85.8 | #992 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=mega-bench-task-red-teaming-visual-misleading · How It Works · Data refreshed daily, snapshot 2026-10-20.