Seeing Culture - Grounding mIoU: leaderboard
Seeing Culture Benchmark (EMNLP 2025) requires segmenting the cultural artifact that justifies each correctly answered question; this board reports overall grounding mIoU as evidence of reasoning.
Metric: Overall mIoU (%). Source: seeingculture-benchmark.github.io. Saturation forecast: Estimated already saturated. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 2.5 VL 7B | 47.2 | #643 |
| 2 | O3 | 31.78 | #124 |
| 3 | Gemini 2.5 Pro | 16.22 | #152 |
| 4 | GPT-4.1 | 13.74 | #248 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=seeing-culture-grounding-miou · How It Works · Data refreshed daily, snapshot 2026-10-20.