OCRRouterAPI-kulcs kérése

Changelog

What changed in the API and the site, newest first.

  1. More models, positions in JSON

    • PaddleOCR-VL, GLM-OCR, Azure Document Intelligence, Google Document AI and Amazon Textract are connected.
    • JSON output carries bounding boxes, confidence scores and page size for engines that report them.
    • DeepSeek-OCR, Jina OCR and Qianfan OCR are connected. DeepSeek-OCR returns positions.
    • GLM-OCR has a second upstream that takes over if the first one fails.
    • Gemini, GPT and Claude are connected.
    • New page: cheapest OCR API, with every model's price per 1,000 pages.

  2. Turn any file into a link

    • Free tools: PDF to link, image to link, video to link, MP3 to link and file to link.
    • POST /v1/files with public=true returns a public URL for any file type, and DELETE /v1/files/{id} removes it.

  3. Comparison pages, pricing calculator, status and API reference

    • Model comparison pages and alternatives pages for every model in the catalog.
    • Pricing calculator: estimate monthly cost by page volume and model.
    • Status page with live availability and the success rate of recent requests.
    • API reference split out from the guides, one section per endpoint.
    • Measured speed and success rate on each model page.

  4. Chinese versions of the site

    • Traditional and Simplified Chinese for the home page, free tools, pricing, model list and sign-up.
    • Language switcher in the footer.

  5. First version of the API

    • POST /v1/ocr with Markdown and JSON output, page ranges, fallbacks and auto routing.
    • File uploads, request history, per-key credit limits and an activity log.
    • Free upload box on the site and temporary keys for agents.
    • CLI and MCP server.