Polish Cultural Vision Benchmark - Country: leaderboard
The same Polish cultural images, asking only which country they belong to. Models score far higher here than on naming the object, which separates recognising a cultural context from actually knowing its contents.
Metric: Accuracy (%). Source: huggingface.co. Saturation forecast: Around December 2027. 42 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Flash | 63.78 | #99 |
| 2 | Gemini 3.1 Pro (Preview) | 63.76 | #58 |
| 3 | Gemini 3.1 Flash Lite | 54.69 | #191 |
| 4 | Seed 2.0 Mini | 51.82 | #276 |
| 5 | Gemini 2.5 Flash | 50.16 | #245 |
| 6 | Gemini 2.5 Pro | 49.7 | #152 |
| 7 | Claude Opus 4.6 | 48.98 | #63 |
| 8 | Gemini 2.5 Flash Lite | 46.34 | #422 |
| 9 | Claude Sonnet 4.6 | 40.42 | #90 |
| 10 | Claude 3.7 Sonnet | 38.87 | #249 |
| 11 | Qwen 3.5 Plus (02-15) | 33.27 | #218 |
| 12 | Grok 4 Fast | 32.77 | #254 |
| 13 | GPT-5.4 | 32.48 | #79 |
| 14 | Qwen 3 VL 235B A22B | 32.47 | #238 |
| 15 | Gemma 3 27B | 30.6 | #607 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=polish-cultural-vision-benchmark-country · How It Works · Data refreshed daily, snapshot 2026-10-20.