--- title: PP-OCRv6 on Hugging Face: 50-Language OCR from 1.5M to... description: PP-OCRv6 is the latest PaddleOCR model family for real-world text detection and recognition, scaling from 1.5M to 34.5M parameters across tiny, small, and... canonical: https://daily.dev/posts/pp-ocrv6-on-hugging-face-50-language-ocr-from-1-5m-to-34-5m-parameters-b5jaxussj twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: PP-OCRv6 on Hugging Face: 50-Language OCR from 1.5M to 34.5M Parameters | daily.dev og:description: PP-OCRv6 is the latest PaddleOCR model family for real-world text detection and recognition, scaling from 1.5M to 34.5M parameters across tiny, small, and... og:url: https://daily.dev/posts/pp-ocrv6-on-hugging-face-50-language-ocr-from-1-5m-to-34-5m-parameters-b5jaxussj og:image: https://api.daily.dev/og/posts/B5JaxUSSj.png og:image:alt: PP-OCRv6 on Hugging Face: 50-Language OCR from 1.5M to 34.5M Parameters og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # PP-OCRv6 on Hugging Face: 50-Language OCR from 1.5M to 34.5M Parameters **[Hugging Face](https://daily.dev/sources/huggingface)** · 4 min read · 0 upvotes · 0 comments ## Summary PP-OCRv6 is the latest PaddleOCR model family for real-world text detection and recognition, scaling from 1.5M to 34.5M parameters across tiny, small, and medium tiers. The medium and small models support 50 languages including Chinese, Japanese, and 46 Latin-script languages. Key architectural improvements include RepLKFPN for multi-scale text detection and EncoderWithLightSVTR for recognition. Compared to PP-OCRv5_server, the medium tier improves detection Hmean by +4.6 points and recognition accuracy by +5.1 points. Models are available on Hugging Face Hub in safetensors, Paddle inference, and ONNX formats, with support for Transformers, ONNX Runtime, and Paddle Inference backends. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What accuracy improvement does PP-OCRv6 offer over PP-OCRv5_server for OCR? PP-OCRv6_medium improves text detection by 4.6 percentage points and text recognition by 5.1 percentage points over PP-OCRv5_server, reaching 86.2% detection Hmean and 83.2% recognition accuracy on PaddleOCR's in-house multi-scenario benchmarks. This makes it a notable accuracy upgrade within the same lightweight OCR model family. _Developers evaluating OCR upgrades can track model benchmarks like this one on daily.dev._ ### How many languages does PP-OCRv6 support for OCR? The medium and small tiers of PP-OCRv6 support 50 languages in a single model family, covering Simplified Chinese, Traditional Chinese, English, Japanese, and 46 Latin-script languages, reducing the need for separate OCR models across multilingual scenarios. _Teams building multilingual document pipelines can follow OCR model updates like this on daily.dev._ ### What inference backends can I use to run PaddleOCR's PP-OCRv6 models? PaddleOCR 3.7 provides a unified inference-engine interface supporting three backends: Paddle Inference (native format, default), a Transformers backend for Hugging Face/PyTorch-oriented inference enabled with engine="transformers", and ONNX Runtime for portable ONNX-based deployment. _Developers choosing a deployment path for OCR models can compare backend options via daily.dev._ ## Similar posts on daily.dev - [PaddleOCR 3.5: Running OCR and Document Parsing Tasks with a Transformers Backend](https://daily.dev/posts/paddleocr-3-5-running-ocr-and-document-parsing-tasks-with-a-transformers-backend-k312uokek) · Hugging Face · 0 upvotes · 0 comments - [Medium](https://daily.dev/posts/medium-kwiiempsu) · Medium · 0 upvotes · 0 comments - [OCR Showdown — Evaluating Three OCR Models](https://daily.dev/posts/ocr-showdown-evaluating-three-ocr-models-ch9qm0xde) · Medium · 0 upvotes · 0 comments - [How Grab Built a Vision LLM to Scan Images](https://daily.dev/posts/how-grab-built-a-vision-llm-to-scan-images-okyf7u3na) · ByteByteGo · 54 upvotes · 0 comments --- Tags: [#machine-learning](https://daily.dev/tags/machine-learning) [View this post on daily.dev](https://daily.dev/posts/pp-ocrv6-on-hugging-face-50-language-ocr-from-1-5m-to-34-5m-parameters-b5jaxussj) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"PP-OCRv6 on Hugging Face: 50-Language OCR from 1.5M to 34.5M Parameters","url":"https://daily.dev/posts/pp-ocrv6-on-hugging-face-50-language-ocr-from-1-5m-to-34-5m-parameters-b5jaxussj","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/pp-ocrv6-on-hugging-face-50-language-ocr-from-1-5m-to-34-5m-parameters-b5jaxussj"},"datePublished":"2026-06-22T13:19:48.812Z","dateModified":"2026-09-13T19:49:12.280Z","description":"PP-OCRv6 is the latest PaddleOCR model family for real-world text detection and recognition, scaling from 1.5M to 34.5M parameters across tiny, small, and...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/bb4bb16457ce5e901cb222890f5a1515?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/bb4bb16457ce5e901cb222890f5a1515?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Hugging Face","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Hugging Face","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/f1f55c67d81a4330acf5b90b26b0c8e1","url":"https://daily.dev/sources/huggingface"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/pp-ocrv6-on-hugging-face-50-language-ocr-from-1-5m-to-34-5m-parameters-b5jaxussj","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"machine-learning","timeRequired":"PT4M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Hugging Face","item":"https://daily.dev/sources/huggingface"},{"@type":"ListItem","position":3,"name":"PP-OCRv6 on Hugging Face: 50-Language OCR from 1.5M to 34.5M Parameters"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/pp-ocrv6-on-hugging-face-50-language-ocr-from-1-5m-to-34-5m-parameters-b5jaxussj#faq","mainEntity":[{"@type":"Question","name":"What accuracy improvement does PP-OCRv6 offer over PP-OCRv5_server for OCR?","acceptedAnswer":{"@type":"Answer","text":"PP-OCRv6_medium improves text detection by 4.6 percentage points and text recognition by 5.1 percentage points over PP-OCRv5_server, reaching 86.2% detection Hmean and 83.2% recognition accuracy on PaddleOCR's in-house multi-scenario benchmarks. This makes it a notable accuracy upgrade within the same lightweight OCR model family. Developers evaluating OCR upgrades can track model benchmarks like this one on daily.dev."}},{"@type":"Question","name":"How many languages does PP-OCRv6 support for OCR?","acceptedAnswer":{"@type":"Answer","text":"The medium and small tiers of PP-OCRv6 support 50 languages in a single model family, covering Simplified Chinese, Traditional Chinese, English, Japanese, and 46 Latin-script languages, reducing the need for separate OCR models across multilingual scenarios. Teams building multilingual document pipelines can follow OCR model updates like this on daily.dev."}},{"@type":"Question","name":"What inference backends can I use to run PaddleOCR's PP-OCRv6 models?","acceptedAnswer":{"@type":"Answer","text":"PaddleOCR 3.7 provides a unified inference-engine interface supporting three backends: Paddle Inference (native format, default), a Transformers backend for Hugging Face/PyTorch-oriented inference enabled with engine=\"transformers\", and ONNX Runtime for portable ONNX-based deployment. Developers choosing a deployment path for OCR models can compare backend options via daily.dev."}}]} ```