Official benchmarks map / Real5-OmniDocBench
PaddlePaddle/PaddleOCR-VL-1.6 is ranked first on the Real5-OmniDocBench leaderboard with 93.19, 1.2% ahead of PaddlePaddle/PaddleOCR-VL-1.5. Read live from the Hugging Face API on November 6, 2026.
| # | Model | Organization | Score |
|---|---|---|---|
| 1 | PaddlePaddle/PaddleOCR-VL-1.6 | PaddlePaddle | 93.19 |
| 2 | PaddlePaddle/PaddleOCR-VL-1.5 | PaddlePaddle | 92.05 |
| 3 | zai-org/GLM-5.3-Flash | zai-org | 90.76 |
| 4 | XingChen-AGI/TeleOCR | XingChen-AGI | 90.72 |
| 5 | zai-org/GLM-OCR | zai-org | 90.32 |
| 6 | moonshotai/Kimi-K2.6 | moonshotai | 89.76 |
| 7 | moonshotai/Kimi-K2.5 | moonshotai | 89.09 |
| 8 | opendatalab/MinerU2.5-Pro-2604-1.2B | opendatalab | 88.94 |
| 9 | Qwen/Qwen3-VL-235B-A22B-Instruct | Qwen | 88.9 |
| 10 | Qwen/Qwen2.5-VL-72B-Instruct | Qwen | 86.92 |
Real5-OmniDocBench is a dataset that Hugging Face marks as an official benchmark. Its leaderboard is built automatically from the .eval_results files that model repositories publish, so entries are mostly self-reported by model authors and evaluation settings differ between submissions. A verified tag appears only where Hugging Face marks an entry as verified.
Original board: huggingface.co/datasets/PaddlePaddle/Real5-OmniDocBench
Part of the Hugging Face official benchmarks map by QuantID. Unofficial. Not affiliated with or endorsed by Hugging Face.