Azure OpenAI cannot receive image_url and file_url as input in responses API

I am currently testing the Azure OpenAI Responses API based on the recommendation to upload files first and reference them using file_id, with the file upload purpose set to assistants.

This approach works correctly when the uploaded file is a PDF. However, when I upload other file types (such as DOCX, XLSX, etc.) and reference their file_id in the request, the API consistently returns the following error:

{
    "error": {
        "message": "The file type you uploaded is not supported. Please try again with a pdf",
        "type": "invalid_request_error",
        "param": "input",
        "code": "unsupported_file"
    }

It appears that Azure OpenAI currently only supports PDF files for input_file usage in the Responses.

I would appreciate clarification on the following points:

1. Is PDF currently the only supported file type for input_file in Azure OpenAI Responses?
2. Are there any plans or timelines to support additional formats such as Word or Excel?
3. Is there a recommended workaround for handling non-PDF documents?

Any clarification or official documentation reference would be appreciated.

Hello Unif,

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

I understand that your Azure OpenAI cannot receive image_url and file_url as input in responses API.

Most of all, both OpenAI and Azure have shown intermittent remote-download issues on Responses for image_url. It’s not uniquely Azure. However, file_url not being supported is an Azure behavior you must adapt to (use Base64 or file_id). The reliable, production-grade path is no remote fetch: send Base64 or upload files first. - https://community.openai.com/t/responses-api-image-upload-error-worked-before/1289584 and https://learn.microsoft.com/en-us/azure/ai-foundry/openai/how-to/responses?view=foundry-classic give more insight.

Therefore, follow this step for Azure enviroment:

1. Python (produces Base64 and calls Responses API): Replace YOUR-RESOURCE-NAME, YOUR-API-KEY, and ensure you deployed a vision model like gpt-4o/gpt-4.1.

      import base64, os
      from openai import OpenAI
      # Azure OpenAI v1 endpoint + API key in Bearer header
      client = OpenAI(
          api_key=os.getenv("AZURE_OPENAI_API_KEY"),
          base_url="https://YOUR-RESOURCE-NAME.openai.azure.com/openai/v1/"
      )
      # Read local image and encode as Base64
      with open("doge.jpg", "rb") as f:
          b64 = base64.b64encode(f.read()).decode("utf-8")
      resp = client.responses.create(
          model="gpt-4.1",  # your deployed model name
          input=[{
              "role": "user",
              "content": [
                  {"type": "input_text", "text": "what is in this image?"},
                  {"type": "input_image", "image_url": f"data:image/jpeg;base64,{b64}"}
              ]
          }]
      )
      print(resp.output_text)
   

   • https://learn.microsoft.com/en-us/azure/ai-foundry/openai/how-to/responses?view=foundry-classic
   If you specifically want curl, supply the Base64 string in the JSON body:

      curl "https://YOUR-RESOURCE-NAME.openai.azure.com/openai/v1/responses?api-version=2025-04-01-preview" \
        -H "Content-Type: application/json" \
        -H "Authorization: Bearer $AZURE_OPENAI_AUTH_TOKEN" \
        -d '{
          "model": "gpt-4.1",
          "input": [{
            "role": "user",
            "content": [
              {"type": "input_text", "text": "what is in this image?"},
              {"type": "input_image", "image_url": "data:image/jpeg;base64,BASE64_STRING_HERE"}
            ]
          }]
        }'
   

2. For PDFs and other files, upload first, then reference file_id (not file_url). Upload the file and the purpose must be assistants for now on Azure:

      curl "https://YOUR-RESOURCE-NAME.openai.azure.com/openai/files?api-version=2025-04-01-preview" \
        -H "Authorization: Bearer $AZURE_OPENAI_AUTH_TOKEN" \
        -F purpose="assistants" \
        -F file="@your_file.pdf"
   

   Then reference it in Responses using input_file.file_id:

      curl "https://YOUR-RESOURCE-NAME.openai.azure.com/openai/v1/responses?api-version=2025-04-01-preview" \
        -H "Content-Type: application/json" \
        -H "Authorization: Bearer $AZURE_OPENAI_AUTH_TOKEN" \
        -d '{
          "model": "gpt-4.1",
          "input": [{
            "role": "user",
            "content": [
              {"type":"input_file","file_id":"<FILE_ID_FROM_UPLOAD>"},
              {"type":"input_text","text":"Summarize this PDF"}
            ]
          }]
        }'
   

   • Check the same link as above for more details.
3. If you must use image_url, then you have to harden it as a last resort by ensuring the URL is publicly reachable over HTTP/HTTPS (no auth, no cookies), and the host doesn’t block bots (e.g., via robots or UA filtering), avoids slow redirects, and returns quickly. - https://learn.microsoft.com/en-us/azure/ai-foundry/openai/how-to/gpt-with-vision?view=foundry-classic, https://community.openai.com/t/responses-api-image-upload-error-worked-before/1289584, and https://community.make.com/t/random-timeout-while-error-when-sending-image-link-to-openai-module/58862 gives more details.
   1. Also, avoid private endpoints/SAS‑restricted blob URLs, the model can’t fetch them. Use Base64 or upload to Files instead.

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.