PP-OCRv6 on Hugging Face: 50-Language OCR from 1.5M to 34.5M Parameters

PaddleOCR has announced the release of PP-OCRv6, the latest generation of its universal optical character recognition model family. Built for real-world text detection and recognition across documents, screenshots, digital displays, industrial labels, and scene text, the family scales from 1.5M to 34.5M parameters across tiny, small, and medium tiers. The small and medium tiers support 50 languages within a single model family, covering Simplified Chinese, Traditional Chinese, English, Japanese, and 46 Latin-script languages. On PaddleOCR's official benchmarks, PP-OCRv6 medium reaches 86.2 percent detection Hmean and 83.2 percent recognition accuracy, marking an improvement over PP-OCRv5_server. PP-OCRv6 utilizes PPLCNetV4 as a unified backbone, upgrades text detection with RepLKFPN for handling multi-scale text, and applies EncoderWithLightSVTR for recognition tasks involving challenging and multilingual crops. The models can output structured JSON and visualization images for downstream use in document parsing, search, extraction, retrieval-augmented generation, and agent workflows. For deployment, PP-OCRv6 supports multiple inference backends through PaddleOCR, including a Transformers backend on the Hugging Face Hub, ONNX variants for ONNX Runtime environments, and safetensors or Paddle inference models.