WikiVQABench: leaderboard
Visual Question Answering (VQA) benchmarks have largely emphasized perception-based tasks that can be solved from visual content alone. In contrast, many real-world scenarios require external knowledge that is not directly observable in the image to answer correctly. We introduce WikiVQABench, a human-curated knowledge-grounded VQA benchmark constructed by systematically combining Wikipedia images
Metric: Accuracy (self-reported). Source: benchmarklist.com. Saturation forecast: Around June 2027. 15 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | InternVL3-78B | 75.6 | #352 |
| 2 | Claude Opus 4.6 | 70.3 | #63 |
| 3 | Claude Sonnet 4.6 | 66.3 | #90 |
| 4 | Llama 3.2 90B Vision Instruct | 65.7 | #622 |
| 5 | Qwen 3 VL 32B Instruct | 64 | #283 |
| 6 | Qwen 3 VL 8B Instruct | 63.1 | #420 |
| 7 | Qwen 3 VL 4B Instruct | 60.2 | #515 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=wikivqabench · How It Works · Data refreshed daily, snapshot 2026-10-20.