PorTEXTO - In-The-Wild: leaderboard
PorTEXTO is a European Portuguese visual-text-extraction benchmark over four subsets (handwritten region and full page, synthetic documents, in-the-wild images). This board reports ANLS (%) on the in-the-wild image subset; higher is better.
Metric: ANLS (%) on the in-the-wild image subset; higher is better. Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 18 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen3.6-27B [28] | 90.5 | |
| 2 | Gemma-4-31B-it [8] | 89.8 | |
| 3 | Qwen3.5-9B [27] | 85.4 | |
| 4 | DeepSeek-OCR-2-3B [33] | 84.1 | |
| 5 | Bee-RL-8B [37] | 83.1 | |
| 6 | Gemma-4-E4B-it [8] | 82.7 | |
| 7 | InternVL3.5-8B [32] | 81 | |
| 8 | LLaVA-OneVision-1.5-8B [1] | 77.4 | |
| 9 | Nemotron-Nano-12B-VL-V2 [5] | 76.8 | |
| 10 | Nemotron-OCR-V2 [14] | 75.9 |
Interactive version: theaggregate.ai/benchmark?slug=portexto-in-the-wild · How It Works · Data refreshed daily, snapshot 2026-11-13.