# OCRRouter > Online OCR for images and PDFs, with an API included. Upload images and PDFs, one at a time or in a batch, and get editable text, tables and data. Export to Word, Excel, Markdown or JSON. OCRRouter picks the OCR model that reads each document best and falls back to another when one fails. Developers call the same service through one API. ## Online OCR - [Online OCR](https://ocrrouter.ai/app): upload files in the browser, batch process, keep a history and export results. - 3 free pages a day without an account; 50 free credits on sign-up. - [Free tools](https://ocrrouter.ai/tools): single-purpose converters such as image to text and PDF to text. ## API (for developers) - Endpoint: POST https://api.ocrrouter.ai/v1/ocr - Auth: Authorization: Bearer - Body: { "model": "|auto", "file_url": "", "format": "markdown|json", "fallbacks": [""], "route": "cheapest|fastest|quality", "document_type": "invoice|receipt|passport|id_card|bank_statement|menu" } - document_type (optional): the response also has "fields", a JSON object with a fixed schema for that type (invoice number, totals, line items; passport number, MRZ and mrz_valid; ...), for 11 extra credits per document. - Pricing: credits per page, $1 = 1000 credits, 50 free credits on sign-up. Failed and cached requests are free. - File to link: POST https://api.ocrrouter.ai/v1/files (multipart: file, public=true) returns a public "url" for any file type. Use it as file_url. - [Docs](https://ocrrouter.ai/docs) - [Pricing](https://ocrrouter.ai/pricing) ## Models - [mistral/mistral-ocr](https://ocrrouter.ai/models/mistral/mistral-ocr): Fast document OCR that returns clean Markdown with tables and embedded images. 25 credits per page. - [google/gemini-flash](https://ocrrouter.ai/models/google/gemini-flash): General vision model with strong handwriting and layout understanding at a low price. 15 credits per page. - [openai/gpt](https://ocrrouter.ai/models/openai/gpt): General vision model. Good at messy scans and following extraction instructions. 11 credits per page. - [anthropic/claude](https://ocrrouter.ai/models/anthropic/claude): General vision model with careful reading of long, dense documents. 25 credits per page. - [llamaindex/llamaparse](https://ocrrouter.ai/models/llamaindex/llamaparse): Document parser built for RAG pipelines, with several quality tiers. 5 credits per page. - [mathpix/convert](https://ocrrouter.ai/models/mathpix/convert): Specialist for math, science papers and handwritten formulas. Outputs LaTeX. 7 credits per page. - [reducto/parse](https://ocrrouter.ai/models/reducto/parse): High-accuracy parsing for complex financial and legal documents, with citations. 20 credits per page. - [datalab/marker](https://ocrrouter.ai/models/datalab/marker): Hosted version of the open-source Marker and Surya models. Strong on books and papers. 14 credits per page. - [aws/textract](https://ocrrouter.ai/models/aws/textract): AWS OCR with form key-value pairs, tables and per-word confidence scores. 25 credits per page. - [azure/document-intelligence](https://ocrrouter.ai/models/azure/document-intelligence): Azure layout model with tables, selection marks and prebuilt invoice and ID readers. 60 credits per page. - [google/document-ai](https://ocrrouter.ai/models/google/document-ai): Google Cloud OCR with layout parsing and specialised processors. 10 credits per page. - [deepseek/deepseek-ocr](https://ocrrouter.ai/models/deepseek/deepseek-ocr): Open-source vision model that compresses pages into few tokens. 1 credits per page. - [jina/jina-ocr](https://ocrrouter.ai/models/jina/jina-ocr): Compact page-to-Markdown model from Jina AI, built on the DeepSeek-OCR encoder. 1 credits per page. - [baidu/qianfan-ocr](https://ocrrouter.ai/models/baidu/qianfan-ocr): Baidu's OCR-tuned vision model. Strong on Chinese documents and tables. 2 credits per page. - [paddle/paddleocr-vl](https://ocrrouter.ai/models/paddle/paddleocr-vl): Open-source document model from Baidu. Excellent on Chinese, Japanese and Korean. 6 credits per page. - [zai/glm-ocr](https://ocrrouter.ai/models/zai/glm-ocr): Small open-source OCR model from Z.ai with layout regions. MIT licensed. 6 credits per page. - [baidu/unlimited-ocr](https://ocrrouter.ai/models/baidu/unlimited-ocr): Open-source model from Baidu that reads hundreds of pages in one pass with flat memory use. 2 credits per page. - [opendatalab/mineru](https://ocrrouter.ai/models/opendatalab/mineru): Open-source PDF to Markdown pipeline popular for academic and Chinese documents. 2 credits per page. - [ibm/docling](https://ocrrouter.ai/models/ibm/docling): IBM's open-source document converter with a structured document model. 2 credits per page. - [tesseract/tesseract](https://ocrrouter.ai/models/tesseract/tesseract): The classic open-source OCR engine. Plain text from clean printed scans. Free. - [paddle/paddleocr](https://ocrrouter.ai/models/paddle/paddleocr): Baidu's classic open-source text detection and recognition pipeline. Light enough to run on CPU. 1 credits per page. - [jaided/easyocr](https://ocrrouter.ai/models/jaided/easyocr): Python OCR library with a simple API. Returns text lines with boxes and confidence. 1 credits per page. - [ocrmypdf/ocrmypdf](https://ocrrouter.ai/models/ocrmypdf/ocrmypdf): Command-line tool that adds a searchable text layer to scanned PDFs, using Tesseract. 1 credits per page. - [rapidai/rapidocr](https://ocrrouter.ai/models/rapidai/rapidocr): PaddleOCR models packaged for ONNX Runtime and other engines. Easy to deploy without Paddle. 1 credits per page. - [mindee/doctr](https://ocrrouter.ai/models/mindee/doctr): Open-source OCR library from Mindee with separate text detection and recognition models. 1 credits per page. - [kraken/kraken](https://ocrrouter.ai/models/kraken/kraken): OCR engine for historical documents and manuscripts, with trainable models for unusual scripts. 1 credits per page. - [microsoft/trocr](https://ocrrouter.ai/models/microsoft/trocr): Transformer text recognition model. Reads single text lines, including handwriting. 1 credits per page. - [allenai/olmocr](https://ocrrouter.ai/models/allenai/olmocr): Open-source vision model from Ai2 (often searched as olm ocr) that turns PDFs into clean, reading-order text. Latest release: olmOCR 2. 2 credits per page. - [datalab/chandra](https://ocrrouter.ai/models/datalab/chandra): Datalab's OCR model that keeps layout and handles forms, tables and handwriting. 2 credits per page. - [datalab/surya](https://ocrrouter.ai/models/datalab/surya): Open-source OCR toolkit with line detection, layout analysis, reading order and tables. 2 credits per page. - [rednote/dots-ocr](https://ocrrouter.ai/models/rednote/dots-ocr): Single vision model from rednote that does layout detection and content recognition together. 2 credits per page. - [nanonets/nanonets-ocr](https://ocrrouter.ai/models/nanonets/nanonets-ocr): Open image-to-Markdown model that tags tables, equations, signatures, watermarks and checkboxes. 2 credits per page. - [tencent/hunyuan-ocr](https://ocrrouter.ai/models/tencent/hunyuan-ocr): Small end-to-end OCR model from Tencent covering parsing, extraction and photo translation. 2 credits per page. - [qwen/qwen-vl](https://ocrrouter.ai/models/qwen/qwen-vl): Alibaba's open-weight vision language models (Qwen2.5-VL, Qwen3-VL). Strong general OCR and grounding. 2 credits per page. - [ibm/granite-docling](https://ocrrouter.ai/models/ibm/granite-docling): Compact IBM vision model that converts pages to DocTags for the Docling pipeline. 1 credits per page. - [huggingface/smoldocling](https://ocrrouter.ai/models/huggingface/smoldocling): Tiny 256M vision model from Hugging Face and IBM that outputs DocTags. Predecessor of Granite Docling. 1 credits per page. - [monkeyocr/monkeyocr](https://ocrrouter.ai/models/monkeyocr/monkeyocr): Open-source document parser that splits work into structure, recognition and relation steps. 2 credits per page. - [meta/nougat](https://ocrrouter.ai/models/meta/nougat): Meta's model that converts academic PDFs to Markdown with LaTeX math. 2 credits per page. - [naver/donut](https://ocrrouter.ai/models/naver/donut): OCR-free document understanding model. Fine-tuned for receipts, forms and document Q&A. 1 credits per page. - [stepfun/got-ocr-2](https://ocrrouter.ai/models/stepfun/got-ocr-2): Unified OCR model that reads plain text, tables, formulas, sheet music and charts. 1 credits per page. - [typhoon/typhoon-ocr](https://ocrrouter.ai/models/typhoon/typhoon-ocr): Open vision model tuned for Thai and English documents, with Markdown output. 2 credits per page. - [microsoft/florence-2](https://ocrrouter.ai/models/microsoft/florence-2): Small Microsoft vision model (0.23B and 0.77B) with OCR and OCR-with-regions tasks that return text with boxes. 1 credits per page. - [tesseract/tesseract-js](https://ocrrouter.ai/models/tesseract/tesseract-js): Tesseract compiled to JavaScript and WebAssembly, so OCR runs in the browser or in Node.js. 1 credits per page. - [fudan/openocr](https://ocrrouter.ai/models/fudan/openocr): Fudan's open toolkit for text detection and recognition, formulas, tables and document parsing. 1 credits per page. - [reducto/rolmocr](https://ocrrouter.ai/models/reducto/rolmocr): Reducto's open PDF-to-text model, built on Qwen2.5-VL-7B with the olmOCR recipe. Needs no PDF metadata and handles rotated pages. 2 credits per page. - [mistral/pixtral](https://ocrrouter.ai/models/mistral/pixtral): Mistral's open-weight vision models, Pixtral 12B and Pixtral Large. Mistral's API no longer serves them, so they run self-hosted. 2 credits per page. - [opengvlab/internvl](https://ocrrouter.ai/models/opengvlab/internvl): Shanghai AI Lab's open vision-language family (1B to 241B, August 2025), strong on OCR and document benchmarks. 2 credits per page. - [openbmb/minicpm-v](https://ocrrouter.ai/models/openbmb/minicpm-v): 8B vision model from OpenBMB (August 2025) built to run on devices, with strong OCR and document parsing. 2 credits per page. - [deepseek/deepseek-vl2](https://ocrrouter.ai/models/deepseek/deepseek-vl2): DeepSeek's mixture-of-experts vision models (December 2024) with OCR and table and chart understanding. 2 credits per page. - [microsoft/phi-4-multimodal](https://ocrrouter.ai/models/microsoft/phi-4-multimodal): Microsoft's 5.6B text, vision and audio model (February 2025). Scores 84.4 on OCRBench per its model card. 2 credits per page. - [google/cloud-vision](https://ocrrouter.ai/models/google/cloud-vision): Google Cloud's image OCR with text and document text detection, word boxes and confidence. 2 credits per page. - [ocrspace/ocr-space](https://ocrrouter.ai/models/ocrspace/ocr-space): Low-cost OCR API with a large free tier. Returns text, word boxes and searchable PDFs. 1 credits per page. - [unstructured/unstructured](https://ocrrouter.ai/models/unstructured/unstructured): Document ETL for RAG. Partitions many file types into typed elements; open-source library too. 20 credits per page. - [chunkr/chunkr](https://ocrrouter.ai/models/chunkr/chunkr): Document parsing API that segments and chunks pages for RAG. Self-hostable version available. 14 credits per page. - [abbyy/finereader](https://ocrrouter.ai/models/abbyy/finereader): Long-standing desktop OCR software. ABBYY also sells OCR SDKs and a cloud API. Price is an estimate. 14 credits per page. - [adobe/acrobat-ocr](https://ocrrouter.ai/models/adobe/acrobat-ocr): OCR built into Acrobat desktop. Adobe PDF Services API also offers OCR. Price is an estimate. 14 credits per page. - [openai/gpt-5](https://ocrrouter.ai/models/openai/gpt-5): OpenAI's flagship model (August 2025). Reads images and PDFs and follows detailed extraction instructions. 14 credits per page. - [openai/gpt-5-mini](https://ocrrouter.ai/models/openai/gpt-5-mini): Smaller, cheaper GPT-5 model that still accepts images and PDFs. 4 credits per page. - [openai/gpt-4-1](https://ocrrouter.ai/models/openai/gpt-4-1): OpenAI model from April 2025 with image input and a 1M-token context window. 11 credits per page. - [openai/gpt-4-1-mini](https://ocrrouter.ai/models/openai/gpt-4-1-mini): Smaller GPT-4.1 model with image input at a lower price. 3 credits per page. - [openai/gpt-4o](https://ocrrouter.ai/models/openai/gpt-4o): OpenAI's omni model (May 2024) with text and image input. 11 credits per page. - [openai/gpt-4o-mini](https://ocrrouter.ai/models/openai/gpt-4o-mini): Small, low-cost GPT-4o model with image input. 3 credits per page. - [openai/o3](https://ocrrouter.ai/models/openai/o3): OpenAI reasoning model that can reason over images, including crops and rotations. 14 credits per page. - [openai/o4-mini](https://ocrrouter.ai/models/openai/o4-mini): Smaller OpenAI reasoning model with image input. 6 credits per page. - [anthropic/claude-opus-4-1](https://ocrrouter.ai/models/anthropic/claude-opus-4-1): Anthropic's most capable Claude 4 model (August 2025). Reads images and PDFs. 27 credits per page. - [anthropic/claude-sonnet-4-5](https://ocrrouter.ai/models/anthropic/claude-sonnet-4-5): Anthropic model from September 2025 with image and PDF input. 14 credits per page. - [anthropic/claude-sonnet-4](https://ocrrouter.ai/models/anthropic/claude-sonnet-4): Claude 4 model from May 2025 with image and PDF input. 14 credits per page. - [anthropic/claude-haiku-4-5](https://ocrrouter.ai/models/anthropic/claude-haiku-4-5): Fast, lower-cost Claude model (October 2025) with image and PDF input. 6 credits per page. - [google/gemini-2-5-pro](https://ocrrouter.ai/models/google/gemini-2-5-pro): Google's most capable Gemini 2.5 model. Reads images and long PDFs natively. 8 credits per page. - [google/gemini-2-5-flash](https://ocrrouter.ai/models/google/gemini-2-5-flash): Fast Gemini 2.5 model with native image and PDF input. 3 credits per page. - [google/gemini-2-5-flash-lite](https://ocrrouter.ai/models/google/gemini-2-5-flash-lite): Lowest-cost Gemini 2.5 model, with image and PDF input. 1 credits per page. - [google/gemini-2-0-flash](https://ocrrouter.ai/models/google/gemini-2-0-flash): Gemini model from early 2025 with image and PDF input. 2 credits per page. - [google/gemma-3](https://ocrrouter.ai/models/google/gemma-3): Google's open-weight model family (March 2025). The 4B, 12B and 27B sizes accept images. 2 credits per page. - [qwen/qwen3-vl](https://ocrrouter.ai/models/qwen/qwen3-vl): Alibaba's Qwen3 vision-language family (2025), with expanded OCR language support. 2 credits per page. - [qwen/qwen2-5-vl-72b](https://ocrrouter.ai/models/qwen/qwen2-5-vl-72b): Largest Qwen2.5-VL model (January 2025), strong on documents, charts and grounding. 3 credits per page. - [qwen/qwen2-5-vl-7b](https://ocrrouter.ai/models/qwen/qwen2-5-vl-7b): 7B Qwen2.5-VL model that fits on a single GPU. 1 credits per page. - [qwen/qwen-vl-max](https://ocrrouter.ai/models/qwen/qwen-vl-max): Alibaba Cloud's hosted top Qwen-VL model. 4 credits per page. - [qwen/qvq-72b](https://ocrrouter.ai/models/qwen/qvq-72b): Qwen's experimental visual reasoning model (December 2024). 3 credits per page. - [meta/llama-4-maverick](https://ocrrouter.ai/models/meta/llama-4-maverick): Meta's mixture-of-experts Llama 4 model (April 2025) with native image input. 2 credits per page. - [meta/llama-4-scout](https://ocrrouter.ai/models/meta/llama-4-scout): Smaller Llama 4 model with image input and a long context window. 2 credits per page. - [meta/llama-3-2-90b-vision](https://ocrrouter.ai/models/meta/llama-3-2-90b-vision): Meta's larger Llama 3.2 vision model (September 2024). 3 credits per page. - [meta/llama-3-2-11b-vision](https://ocrrouter.ai/models/meta/llama-3-2-11b-vision): Meta's smaller Llama 3.2 vision model (September 2024). 1 credits per page. - [mistral/mistral-medium-3](https://ocrrouter.ai/models/mistral/mistral-medium-3): Mistral's multimodal model from May 2025 with image input. 4 credits per page. - [mistral/mistral-small-3-2](https://ocrrouter.ai/models/mistral/mistral-small-3-2): Open-weight 24B Mistral model with image input. 1 credits per page. - [xai/grok-4](https://ocrrouter.ai/models/xai/grok-4): xAI's flagship model (July 2025) with image input. 14 credits per page. - [xai/grok-2-vision](https://ocrrouter.ai/models/xai/grok-2-vision): xAI's earlier vision model for image understanding. 6 credits per page. - [amazon/nova-pro](https://ocrrouter.ai/models/amazon/nova-pro): Amazon's multimodal model on Bedrock with image and document input. 3 credits per page. - [amazon/nova-lite](https://ocrrouter.ai/models/amazon/nova-lite): Low-cost Amazon Nova model with image and document input. 1 credits per page. - [microsoft/phi-3-5-vision](https://ocrrouter.ai/models/microsoft/phi-3-5-vision): Microsoft's 4.2B vision model (August 2024). 1 credits per page. - [moonshot/kimi-vl](https://ocrrouter.ai/models/moonshot/kimi-vl): Moonshot's mixture-of-experts vision model with about 3B active parameters (April 2025). 1 credits per page. - [zai/glm-4-5v](https://ocrrouter.ai/models/zai/glm-4-5v): Z.ai's open vision-language model (August 2025) built on GLM-4.5-Air. 3 credits per page. - [zai/glm-4-1v-thinking](https://ocrrouter.ai/models/zai/glm-4-1v-thinking): 9B vision model from Z.ai with step-by-step reasoning (July 2025). 1 credits per page. - [bytedance/doubao-seed-1-6](https://ocrrouter.ai/models/bytedance/doubao-seed-1-6): ByteDance's multimodal Doubao model on Volcano Engine, with image input. 3 credits per page. - [huggingface/smolvlm](https://ocrrouter.ai/models/huggingface/smolvlm): Hugging Face's small open vision models (256M to 2.2B). 1 credits per page. - [huggingface/idefics3](https://ocrrouter.ai/models/huggingface/idefics3): Hugging Face's 8B open vision model built on Llama 3.1. 1 credits per page. - [allenai/molmo](https://ocrrouter.ai/models/allenai/molmo): Ai2's open vision-language models (September 2024), with pointing at image regions. 1 credits per page. - [nvidia/llama-nemotron-nano-vl](https://ocrrouter.ai/models/nvidia/llama-nemotron-nano-vl): NVIDIA's 8B vision model tuned for document understanding (June 2025). 1 credits per page. - [llava/llava-onevision](https://ocrrouter.ai/models/llava/llava-onevision): Open vision model from the LLaVA project for single images, multiple images and video. 1 credits per page. - [cohere/aya-vision](https://ocrrouter.ai/models/cohere/aya-vision): Cohere's open-weight multilingual vision models (8B and 32B, March 2025). 1 credits per page. - [cohere/command-a-vision](https://ocrrouter.ai/models/cohere/command-a-vision): Cohere's enterprise vision model (July 2025) for documents, charts and scans. 4 credits per page. - [deepseek/janus-pro](https://ocrrouter.ai/models/deepseek/janus-pro): DeepSeek's unified image understanding and generation model (January 2025). 1 credits per page. - [reka/reka-flash](https://ocrrouter.ai/models/reka/reka-flash): Reka's multimodal model with image, video and audio input. 3 credits per page. - [baidu/ernie-4-5-vl](https://ocrrouter.ai/models/baidu/ernie-4-5-vl): Baidu's open-weight multimodal ERNIE 4.5 models (June 2025). 2 credits per page. ## Free tools - [PDF to Markdown](https://ocrrouter.ai/tools/pdf-to-markdown): Upload a PDF and get clean Markdown with headings, tables and formulas. Works on scanned PDFs too. - [PDF to text](https://ocrrouter.ai/tools/pdf-to-text): Extract plain text from any PDF, including scanned and image-only files. - [Image to text](https://ocrrouter.ai/tools/image-to-text): Extract text from photos, screenshots and scans. JPG, PNG and WebP are supported. - [Handwriting to text](https://ocrrouter.ai/tools/handwriting-to-text): Turn handwritten notes, letters and forms into editable text. Cursive works too. - [PDF to JSON](https://ocrrouter.ai/tools/pdf-to-json): Get every heading, paragraph and table in a PDF as structured JSON with page positions. - [JPG to text](https://ocrrouter.ai/tools/jpg-to-text): Upload a JPG and get the text in it as editable text. Photos, scans saved as JPG and phone pictures all work. - [PNG to text](https://ocrrouter.ai/tools/png-to-text): Extract text from PNG images, including screenshots, exported slides and diagrams with labels. - [Photo to text](https://ocrrouter.ai/tools/photo-to-text): Take a photo of a page, sign or whiteboard with your phone and turn it into text you can edit. - [Picture to text](https://ocrrouter.ai/tools/picture-to-text): Turn any picture with words in it into text. Drop in a picture, get the words back, ready to copy. - [Screenshot to text](https://ocrrouter.ai/tools/screenshot-to-text): Paste a screenshot and get its text. Useful for error messages, chats, code and anything you cannot select. - [GIF to text](https://ocrrouter.ai/tools/gif-to-text): Extract text from GIF images, such as old scans, web graphics and reaction GIFs with captions. - [Extract text from image](https://ocrrouter.ai/tools/extract-text-from-image): Pull all the text out of an image, keeping paragraphs, lists and tables in order, ready to reuse. - [Copy text from image](https://ocrrouter.ai/tools/copy-text-from-image): Can't select the text in an image? Paste a screenshot with Ctrl+V or upload a photo, and copy the text with one click. - [Extract text from PDF](https://ocrrouter.ai/tools/extract-text-from-pdf): Get the text out of any PDF, digital or scanned, in reading order with tables kept as tables. - [OCR PDF](https://ocrrouter.ai/tools/ocr-pdf): Run OCR on a PDF and get its text back. Scanned, faxed and photographed PDFs become readable text. - [Scanned PDF to text](https://ocrrouter.ai/tools/scanned-pdf-to-text): Turn a scanned PDF, where every page is an image, into real text you can edit and search. - [Image to LaTeX](https://ocrrouter.ai/tools/image-to-latex): Upload a picture of an equation and get LaTeX you can paste into Overleaf, a paper or Markdown. - [PDF to LaTeX](https://ocrrouter.ai/tools/pdf-to-latex): Get the text and equations from a PDF with every formula as LaTeX. Good for papers, lecture notes and problem sets. - [Image to Markdown](https://ocrrouter.ai/tools/image-to-markdown): Turn an image of a document into Markdown with headings, lists, tables and LaTeX maths. - [Image to JSON](https://ocrrouter.ai/tools/image-to-json): Get the text in an image as structured JSON: typed blocks with positions and confidence, ready for code. - [Image to table](https://ocrrouter.ai/tools/image-to-table): Upload a picture of a table and get it back with rows and columns intact: copy it as a Markdown table or download an Excel file. - [PDF to Excel](https://ocrrouter.ai/tools/pdf-to-excel): Get the tables out of a PDF, including scanned ones, as an Excel file with one sheet per table. - [PDF to CSV](https://ocrrouter.ai/tools/pdf-to-csv): Pull the tables out of a PDF, scanned ones included, and download them as a CSV file. - [Menu OCR](https://ocrrouter.ai/tools/menu-ocr): Scan a menu photo or PDF to JSON: every dish with its section, description, price and options, ready for a delivery app or POS. - [Image to Excel](https://ocrrouter.ai/tools/image-to-excel): Photograph or screenshot a table and get it back as an Excel file with real rows and columns. - [Image to Word](https://ocrrouter.ai/tools/image-to-word): Turn an image of a document into an editable Word file, with headings, lists and tables. - [JPG to Word](https://ocrrouter.ai/tools/jpg-to-word): Upload a JPG of a letter, form or page and download an editable Word file with the text, headings and tables. - [JPG to Excel](https://ocrrouter.ai/tools/jpg-to-excel): Upload a JPG of a table, receipt or price list and download an Excel file with the rows and columns in place. - [PDF to Word (OCR)](https://ocrrouter.ai/tools/pdf-to-word): Convert scanned and image-only PDFs into an editable Word file, with headings, lists and tables. - [Make PDF searchable](https://ocrrouter.ai/tools/make-pdf-searchable): Find out what text is in a scanned PDF. Today you get the recognised text; a searchable PDF download is coming soon. - [PDF to HTML](https://ocrrouter.ai/tools/pdf-to-html): Turn a PDF, including a scanned one, into clean HTML with real headings, paragraphs and tables. Download the .html file or copy the Markdown. - [Bank statement converter](https://ocrrouter.ai/tools/bank-statement-converter): Read PDF and scanned bank statements and get the transaction tables as an Excel file, or as JSON. - [PDF to TXT](https://ocrrouter.ai/tools/pdf-to-txt): Get the text of a PDF as a .txt file. Scanned and image-only PDFs work too, because every page goes through OCR. - [PDF to Google Docs](https://ocrrouter.ai/tools/pdf-to-google-docs): Get an editable copy of any PDF, scanned or not, ready to paste into Google Docs with headings, lists and tables. - [PDF to link](https://ocrrouter.ai/tools/pdf-to-link): Upload a PDF and get a URL you can paste into a chat, an email or a web page. It opens in the browser, no download needed. - [Image to link](https://ocrrouter.ai/tools/image-to-link): Upload a JPG, PNG, WebP or GIF and get a direct image URL. Use it in a chat, a forum, Markdown or HTML. - [GIF to link](https://ocrrouter.ai/tools/gif-to-link): Upload a GIF, animated or not, and get a direct URL that plays in chats, Markdown, HTML and GitHub. - [Video to link](https://ocrrouter.ai/tools/video-to-link): Upload a short MP4 or WebM clip and get a link that plays in the browser. Video links need a free account. - [MP4 to link](https://ocrrouter.ai/tools/mp4-to-link): Upload an MP4 and get a link that plays in the browser, with seeking. Free with an account. - [MP3 to link](https://ocrrouter.ai/tools/mp3-to-link): Upload an audio file and get a link that plays in the browser. MP3, WAV, M4A and OGG are supported. - [File to link](https://ocrrouter.ai/tools/file-to-link): Upload a document, image, archive or any other file and get a URL for it. No account needed. ## OCR by document and language - [Invoice OCR](https://ocrrouter.ai/ocr/invoice): Turn PDF and scanned invoices into line items, totals and supplier details you can post straight to your accounting system. - [Receipt OCR](https://ocrrouter.ai/ocr/receipt): Read crumpled, faded and photographed receipts and get merchant, date, items and totals back as JSON. - [Passport OCR](https://ocrrouter.ai/ocr/passport): Extract the machine-readable zone and the printed fields from passport photos and scans for KYC and onboarding flows. - [Bank statement OCR](https://ocrrouter.ai/ocr/bank-statement): Convert bank statement PDFs into clean transaction rows for lending, bookkeeping and reconciliation. - [ID card OCR](https://ocrrouter.ai/ocr/id-card): Read national ID cards and driver licences from a phone photo and return the printed fields. - [Resume OCR](https://ocrrouter.ai/ocr/resume): Parse resumes and CVs in any layout into structured work history, education and skills. - [Table OCR](https://ocrrouter.ai/ocr/table): Pull tables out of PDFs and images with merged cells and multi-row headers intact, as Markdown or as rows. - [Contract OCR](https://ocrrouter.ai/ocr/contract): Convert long scanned contracts to clean Markdown with headings and clause numbering preserved, ready for review or search. - [Handwriting OCR](https://ocrrouter.ai/ocr/handwriting): Read handwritten notes, forms and letters, including cursive, and get editable text back. - [Math OCR](https://ocrrouter.ai/ocr/math): Convert printed and handwritten equations, worksheets and papers to LaTeX inside Markdown. - [Business card OCR](https://ocrrouter.ai/ocr/business-card): Turn photos of business cards into contact records with name, title, company, phone and email ready for your CRM. - [Driver's license OCR](https://ocrrouter.ai/ocr/drivers-license): Extract name, license number, dates and address from driver's license photos for KYC, car rental and onboarding. - [Payslip OCR](https://ocrrouter.ai/ocr/payslip): Read payslips and pay stubs into gross pay, deductions and net pay for lending, renting and payroll checks. - [Purchase order OCR](https://ocrrouter.ai/ocr/purchase-order): Extract PO numbers, buyer, ship-to and line items from purchase orders so they can be matched against invoices. - [Bill of lading OCR](https://ocrrouter.ai/ocr/bill-of-lading): Read bills of lading into shipper, consignee, vessel, container and cargo details for freight and customs workflows. - [Medical record OCR](https://ocrrouter.ai/ocr/medical-record): Turn scanned medical records, discharge summaries and lab reports into structured patient, diagnosis and medication data. - [Birth certificate OCR](https://ocrrouter.ai/ocr/birth-certificate): Extract names, dates, places and parents from birth certificates, including older and handwritten registrations. - [Utility bill OCR](https://ocrrouter.ai/ocr/utility-bill): Read electricity, gas, water and internet bills for proof of address, usage tracking and expense reporting. - [Shipping label OCR](https://ocrrouter.ai/ocr/shipping-label): Read shipping labels from parcel photos into tracking number, carrier, sender and recipient addresses. - [License plate OCR](https://ocrrouter.ai/ocr/license-plate): Read license plate numbers from vehicle photos for parking, access control and fleet logs. - [Delivery note OCR](https://ocrrouter.ai/ocr/delivery-note): Extract items, quantities and signatures from delivery notes and packing slips to match goods received against orders. - [Form OCR](https://ocrrouter.ai/ocr/form): Turn filled-in paper and PDF forms into key-value pairs, including checkboxes and handwritten answers. - [Check OCR](https://ocrrouter.ai/ocr/check): Read bank checks (cheques) into payee, amount, date and MICR line for remote deposit and reconciliation. - [Menu OCR](https://ocrrouter.ai/ocr/menu): Turn restaurant menu photos and PDFs into dishes, descriptions, prices and sections for delivery apps and listings. - [Insurance claim OCR](https://ocrrouter.ai/ocr/insurance-claim): Extract policy, claimant, incident and amount details from insurance claim forms and supporting documents. - [W-2 form OCR](https://ocrrouter.ai/ocr/w2): Read US W-2 wage and tax statements into each numbered box for tax prep, lending and income verification. - [Tax form OCR](https://ocrrouter.ai/ocr/tax-form): Extract fields from tax returns and tax forms, such as 1099s, 1040s and VAT returns, into structured data. - [Prescription OCR](https://ocrrouter.ai/ocr/prescription): Read printed and handwritten prescriptions into drug, dose, frequency and prescriber details. - [Certificate OCR](https://ocrrouter.ai/ocr/certificate): Extract names, titles, issuers and dates from diplomas, training certificates and awards for verification. - [Academic paper OCR](https://ocrrouter.ai/ocr/academic-paper): Convert research papers into clean Markdown with headings, equations, tables, figures and references in reading order. - [Book OCR](https://ocrrouter.ai/ocr/book): Turn scanned books into clean, searchable text or Markdown, page by page, with headers and page numbers removed. - [Newspaper OCR](https://ocrrouter.ai/ocr/newspaper): Read newspaper pages and archives into separate articles with headlines, bylines and body text in the right order. - [KYC document OCR](https://ocrrouter.ai/ocr/kyc): Read passports, ID cards, driver's licenses and proof-of-address letters in one KYC flow, and get the identity fields back in one shape whatever the document. - [Bookkeeping document OCR](https://ocrrouter.ai/ocr/bookkeeping): Turn the shoebox of receipts, supplier bills and statements a client sends each month into dated, categorised rows you can import into QuickBooks, Xero or a spreadsheet. - [MRZ OCR](https://ocrrouter.ai/ocr/mrz): Read the two or three lines of chevrons at the bottom of passports, ID cards and visas, parse them into fields and check every check digit. - [Financial statement OCR](https://ocrrouter.ai/ocr/financial-statement): Pull income statements, balance sheets and cash-flow statements out of annual reports and scanned accounts into rows a credit analyst or auditor can model. - [Legal document OCR](https://ocrrouter.ai/ocr/legal-document): Make scanned pleadings, court orders, deeds and discovery productions searchable, with case numbers, parties and dates pulled out for your matter list. - [Credit card statement OCR](https://ocrrouter.ai/ocr/credit-card-statement): Convert credit card statement PDFs into a spreadsheet of transactions with dates, merchants, amounts and card numbers, ready for Excel or your expense tool. - [Insurance card OCR](https://ocrrouter.ai/ocr/insurance-card): Read the front and back of health insurance cards at check-in and get member ID, group number, payer and copays without retyping. - [Customs declaration OCR](https://ocrrouter.ai/ocr/customs-declaration): Extract declarant, HS codes, values, origin and duties from customs declarations and entry summaries such as CBP 7501 and EU SAD forms. - [Commercial invoice OCR](https://ocrrouter.ai/ocr/commercial-invoice): Read export commercial invoices into seller, buyer, Incoterms, HS codes, country of origin and line values for customs filing and landed-cost checks. - [Timesheet OCR](https://ocrrouter.ai/ocr/timesheet): Turn handwritten and printed timesheets from crews and temps into rows of worker, date, start, end, breaks and hours for payroll. - [Packing list OCR](https://ocrrouter.ai/ocr/packing-list): Extract cartons, quantities, weights and dimensions from packing lists so you can check a shipment against the commercial invoice and PO. - [Mortgage document OCR](https://ocrrouter.ai/ocr/mortgage-document): Classify and read loan packages, including the Loan Estimate, Closing Disclosure, note and deed of trust, and pull out the loan terms underwriters check. - [Chinese OCR](https://ocrrouter.ai/ocr/chinese): Read Simplified and Traditional Chinese from scans, screenshots and photos, including vertical text. - [Japanese OCR](https://ocrrouter.ai/ocr/japanese): Read Japanese documents with kanji, kana and furigana, in horizontal or vertical layout. - [Korean OCR](https://ocrrouter.ai/ocr/korean): Read Korean printed and handwritten text, including documents with Hanja. - [Arabic OCR](https://ocrrouter.ai/ocr/arabic): Read Arabic right-to-left text with correct letter joining and diacritics. - [Vietnamese OCR](https://ocrrouter.ai/ocr/vietnamese): Read Vietnamese with every tone mark in place, from IDs, invoices and scanned books. - [Hindi OCR](https://ocrrouter.ai/ocr/hindi): Read Hindi and other Devanagari text, including conjunct characters. - [Thai OCR](https://ocrrouter.ai/ocr/thai): Read Thai text, which has no spaces between words and vowels above and below the line. - [Russian OCR](https://ocrrouter.ai/ocr/russian): Read Russian and other Cyrillic text from documents, including handwriting. - [Turkish OCR](https://ocrrouter.ai/ocr/turkish): Read Turkish with dotted and dotless i, ğ, ş and ç kept exactly right. - [Hebrew OCR](https://ocrrouter.ai/ocr/hebrew): Read Hebrew right-to-left text, with or without vowel points. - [German OCR](https://ocrrouter.ai/ocr/german): Read German documents, including older Fraktur prints, with umlauts and ß preserved. - [Spanish OCR](https://ocrrouter.ai/ocr/spanish): Read Spanish documents with accents, ñ and inverted punctuation kept intact. - [Sanskrit OCR](https://ocrrouter.ai/ocr/sanskrit): Read Sanskrit texts in Devanagari, from printed editions to manuscript scans, with conjuncts and diacritics intact. - [Tamil OCR](https://ocrrouter.ai/ocr/tamil): Read Tamil text from documents, signboards and books, with vowel signs attached to the right consonants. - [Persian OCR](https://ocrrouter.ai/ocr/persian): Read Persian (Farsi) right-to-left text with the extra letters and zero-width joins it uses. - [Urdu OCR](https://ocrrouter.ai/ocr/urdu): Read Urdu text, including Nastaliq print, where letters slope and stack along each word. - [Nepali OCR](https://ocrrouter.ai/ocr/nepali): Read Nepali Devanagari text from government forms, IDs and printed documents. - [Khmer OCR](https://ocrrouter.ai/ocr/khmer): Read Khmer text with stacked subscript consonants and vowels around the base letter. - [Bengali OCR](https://ocrrouter.ai/ocr/bengali): Read Bengali (Bangla) text from books, forms and IDs, with conjuncts and vowel signs kept intact. - [Amharic OCR](https://ocrrouter.ai/ocr/amharic): Read Amharic in Ge'ez script, where each of 200+ characters is a consonant-vowel syllable. - [Telugu OCR](https://ocrrouter.ai/ocr/telugu): Read Telugu text with its rounded letters, vowel signs and stacked consonants. - [Greek OCR](https://ocrrouter.ai/ocr/greek): Read modern Greek text with accents kept, and older polytonic Greek from historical prints. - [French OCR](https://ocrrouter.ai/ocr/french): Read French documents with é, è, ç and other accents kept intact, from invoices to old books. - [Portuguese OCR](https://ocrrouter.ai/ocr/portuguese): Read Brazilian and European Portuguese with ã, õ, ç and accents preserved. - [Italian OCR](https://ocrrouter.ai/ocr/italian): Read Italian documents with accented vowels and apostrophes kept exactly right. - [Polish OCR](https://ocrrouter.ai/ocr/polish): Read Polish text with ą, ę, ł, ż, ź and other diacritics kept intact. - [Indonesian OCR](https://ocrrouter.ai/ocr/indonesian): Read Indonesian documents, IDs and receipts in plain Latin script. - [Ukrainian OCR](https://ocrrouter.ai/ocr/ukrainian): Read Ukrainian Cyrillic text without confusing it for Russian, keeping і, ї, є and ґ. - [Dutch OCR](https://ocrrouter.ai/ocr/dutch): Read Dutch documents with ij, diaereses and long compound words handled correctly. - [Latin OCR](https://ocrrouter.ai/ocr/latin): Read Latin texts from early printed books, manuscripts and inscriptions. - [Burmese OCR](https://ocrrouter.ai/ocr/burmese): Read Burmese text in its round Myanmar script, where vowel signs and medials surround each consonant. - [Mongolian OCR](https://ocrrouter.ai/ocr/mongolian): Read Mongolian in Cyrillic, and traditional vertical Mongolian script where models support it. ## Guides - [Best OCR API in 2026: 13 APIs compared on price and real tests](https://ocrrouter.ai/blog/best-ocr-api): 13 OCR APIs compared: price per 1,000 pages, tables, handwriting and forms, with results from our own tests. Plus free OCR APIs and the best OCR SDK options. - [OCR on iOS: Apple's Vision framework or a cloud OCR API](https://ocrrouter.ai/blog/ios-ocr-api): Two ways to add OCR to an iPhone app: the on-device Vision framework, and a cloud OCR API called from Swift. Code for both, and when to use each. - [Copy text from an image on iPhone, Android, Windows and Mac](https://ocrrouter.ai/blog/copy-text-from-image): Copy text from any image with tools already on your device: Live Text, Google Lens, Snipping Tool, PowerToys. Steps for each, and what to use for full pages. - [Python OCR: pytesseract, EasyOCR and PaddleOCR, tested](https://ocrrouter.ai/blog/python-ocr): OCR in Python with four options, each run on the same invoice: pytesseract, EasyOCR, PaddleOCR 3 and an OCR API. Code, real output and which to pick. - [Extract tables from PDF with Python, even scanned PDFs](https://ocrrouter.ai/blog/extract-tables-from-pdf): Extract tables from digital PDFs with pdfplumber, Camelot or PyMuPDF, and from scanned PDFs with OCR. Tested code, real output and the traps we hit. - [Invoice data extraction: how to do it reliably](https://ocrrouter.ai/blog/invoice-data-extraction): Extract invoice number, dates, line items and totals from invoices with OCR and Python, then catch errors with arithmetic checks. Tested on 8 OCR models. - [Handwritten text recognition: best OCR for handwriting](https://ocrrouter.ai/blog/handwritten-text-recognition): We ran a handwritten note through 10 OCR engines. The best read 97% to 99% of the words; Tesseract read 6%. Results, code and how to choose. - [Best OCR models in 2026: tested picks by use case](https://ocrrouter.ai/blog/best-ocr-model): Best OCR models in 2026: PaddleOCR-VL and GLM-OCR for documents, Mistral OCR for handwriting and low cost, Azure for IDs and forms. Scores, prices, tests. - [OCR accuracy rate: how to measure and improve it (tested)](https://ocrrouter.ai/blog/improve-ocr-accuracy): We damaged an invoice in 6 ways and measured Tesseract before and after each fix. Resolution and deskew mattered most; binarisation did not. Code and checklist.