I wanted to know list of files types supported by each model in Azure Open AI

Question

As we are working with Azure Open AI. We have a requirement to identify the handwritten documents, foreign language, handwritten instructions on the document, if there is any signature present on the document. As we will support different file types and Open AI supports different file types.
We wanted to Azure Open AI supported file types.

Answer 1

Hello

Welcome to the Microsoft Q&A and thank you for posting your questions here.

I understand that you would like to know the list of file types supported by each model in Azure Open AI.

Regarding your questions:

1. Azure OpenAI models like GPT-4 Turbo and GPT-4o with Vision do not directly support document formats such as .pdf or .docx. Instead, they process images embedded in prompts using Markdown syntax (!image). Supported image formats include .jpg, .jpeg, .png, .bmp, .tiff, and .heif. - https://learn.microsoft.com/en-us/azure/machine-learning/prompt-flow/tools-reference/azure-open-ai-gpt-4v-tool?view=azureml-api-2 For document analysis tasks like signature detection, handwriting interpretation, and foreign language understanding, you must first convert documents to image format before passing them to these models. For example, scanned forms or handwritten notes in .pdf must be rendered as images before input. See GPT-4 Vision documentation for details. https://github.com/Azure-Samples/azure-openai-gpt-4-vision-pdf-extraction-sample
2. Azure AI Document Intelligence (formerly Form Recognizer) is the recommended service for structured document analysis. It supports a wide range of formats including .pdf, .docx, .xlsx, .pptx, .html, and image formats like .jpg, .png, .tiff, and .bmp. This service can extract key-value pairs, tables, selection marks, and detect handwritten content and signatures using custom neural models. It handles up to 2,000 pages per document and supports files up to 500 MB in the paid tier. For full capabilities and limits, refer to Document Intelligence v4.0 documentation. https://learn.microsoft.com/en-us/azure/ai-services/document-intelligence/train/custom-neural?view=doc-intel-4.0.0
3. Azure AI Language service is ideal for summarization, classification, and PII redaction of native text-based documents. It supports .txt, .pdf, and .docx formats but does not process scanned or image-embedded text. This service is best used after OCR preprocessing if the document is not natively digital. It supports up to 20 documents per request with a total size limit of 10 MB. See Language service documentation for supported formats and use cases. - https://learn.microsoft.com/en-us/azure/ai-services/language-service/summarization/how-to/document-summarization
4. For a complete workflow: upload your documents to Azure Blob Storage, use Document Intelligence to extract structured data and layout, then pass the extracted content or converted images to GPT-4 Turbo/GPT-4o for reasoning, summarization, or classification. If working with native .pdf or .docx, use Azure AI Language for summarization or redaction. This modular pipeline ensures accurate handling of handwritten notes, foreign languages, and signatures across diverse file types.

I hope this is helpful! Do not hesitate to let me know if you have any other questions or clarifications.

---

Please don't forget to close up the thread here by upvoting and accept it as an answer if it is helpful.

Answer 2

Hello,

Welcome to Microsoft Q&A,

1. Azure OpenAI Assistants API (incl. File Search)
   1. Used when building AI assistants that ingest and interact with uploaded files.
   2. 1. https://learn.microsoft.com/en-us/azure/ai-foundry/openai/how-to/assistant
2. Azure OpenAI On Your Data (Preview)
   1. Designed for ingesting enterprise documents into copilot-like experiences.
   2. Supported formats: .txt, .md, .html, .docx, .pptx, .pdf
   3. https://learn.microsoft.com/en-us/azure/ai-foundry/openai/how-to/file-search?tabs=python
3. Azure AI Document Intelligence (formerly Form Recognizer)
   1. Ideal for analyzing document structure, extracting text (including handwriting via OCR), tables, selection marks, signatures, and more.
   2. Supported Types:
      1. PDF
      2. Images: JPEG/JPG, PNG, BMP, TIFF, HEIF
      3. Office Documents: DOCX, XLSX, PPTX, HTML
   3. https://learn.microsoft.com/en-us/azure/ai-services/document-intelligence/prebuilt/layout?view=doc-intel-4.0.0&tabs=rest%2Csample-code

Please upvote and accept the answer if it helps!!