MEGA-Bench Task - AV Human Multiview Counting: leaderboard
The "AV Human Multiview Counting" 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: Not forecast. 44 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Pixtral-12B | 33.3 | #808 |
| 2 | Gemini 1.5 Flash (002) | 26.7 | #583 |
| 3 | Qwen 2 VL 7B | 26.7 | #750 |
| 4 | Gemma 3 12B (IT) | 20 | #664 |
| 5 | Aquila-VL-2B | 20 | #1238 |
| 6 | GPT-4o | 13.3 | #343 |
| 7 | GPT-4o Mini | 13.3 | #599 |
| 8 | InternVL3-8B | 13.3 | #620 |
| 9 | InternVL3-38B | 13.3 | #405 |
| 10 | Gemini 2.5 Pro (Preview 03-25) | 13.3 | #123 |
| 11 | MiniCPM-V-2.6 | 13.3 | #836 |
| 12 | InternVL3-14B | 13.3 | #505 |
| 13 | GLM-4.1V-9B (Thinking) | 13.3 | #472 (GLM-4.1V-9B) |
| 14 | Llama 4 Scout Base | 13.3 | #707 |
| 15 | Llama 3.2 11B | 13.3 | #1199 |
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-av-human-multiview-counting · How It Works · Data refreshed daily, snapshot 2026-10-20.