GlotOCR Bench - Mid-Resource Scripts: leaderboard

Multilingual OCR on GlotOCR Bench (2026): sentences from real multilingual text in 158 Unicode scripts are rendered as clean images and transcribed zero-shot. Share of sentences with character error rate at most 5 percent, macro-averaged over nine mid-resource scripts such as Arabic, Cyrillic and Devanagari; higher is better.

Metric: Acc@5 (%): share of sentences transcribed with CER at most 0.05, macro-averaged over scripts, mid-resource tier (9 scripts such as Arabic, Cyrillic, Devanagari), on GlotOCR Bench (16,375 real multilingual sentences rendered as clean images across 158 Unicode scripts; zero-shot transcription; CER is the character edit distance with whitespace ignored, capped at 1, best of four output normalizations); higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 14 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Gemini 3.1 Flash Lite82.7#377
2dots.ocr78.3
3Dots.mocr78.1
4HunyuanOCR73.9
5Qwen 3 VL 8B67.1#927
6GPT-4.166.7#506
7olmOCR-263.8
8RolmOCR61.8
9LightOnOCR-251.6
10Nanonets-OCR251.1
11PaddleOCR-VL-1.548.6
12FireRed-OCR46.5
13DeepSeek-OCR-239.7
14GLM-OCR29.8

No result here: #2 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=glotocr-bench-mid-resource-scripts · How It Works · Data refreshed daily, snapshot 2026-11-22.