File inputs using Base64-encoded files not working

Also when I used the completions API I get

AuthenticationError: Error code: 401 - {'error': {'message': 'You must be logged in to use this feature.', 'type': 'invalid_request_error', 'param': None, 'code': None}}

the code to reproduce it.

completion = client.chat.completions.create(
    model="gpt-4.1-mini",
    messages=[
        {
            "role": "user",
            "content": [
                {
                    "type": "file",
                    "file": {
                        "filename": "example.pdf",
                        "file_data": f"data:application/pdf;base64,{base64_string}",
                    },
                },
                {
                    "type": "text",
                    "text": "What is this document about?",
                },
            ],
        },
    ],
)

I get the same Error while trying to upload a base64 encoded PDF to my deployment of GPT4.1 in SwedenCentral.

In addition to that I also get an error when referencing a file by file_id (uploaded via the azure ai foundry menu).

Both methods are documented here https://platform.openai.com/docs/guides/pdf-files?api-mode=chat#base64-encoded-files

When can we expect to use this new feature?

Hi Ali Beyram

If you're experiencing PDF upload errors with the Azure OpenAI API, the issue may stem from server side restrictions, network instability, file size limitations, or content-related security policies. To mitigate transient errors, implementing retry logic with exponential backoff can help ensure smoother uploads by gradually increasing the wait time between retries.
Additionally, verifying that your PDF file adheres to Azure’s 512MB file size limit is crucial, as exceeding this limit may lead to processing failures. If security filters are triggering errors, reviewing the file’s content for restricted elements can help bypass unnecessary blocks. Another effective solution is switching to a different Azure region to rule out local service disruptions. For exceptionally large files, chunked uploads can prevent timeout errors and enhance reliability.

import base64
import os
import time
from openai import AzureOpenAI

client = AzureOpenAI(
    api_version="2025-03-01-preview",
    azure_endpoint=os.environ["AZURE_OPENAI_ENDPOINT"],
    api_key=os.getenv("AZURE_OPENAI_API_KEY"),
)

def upload_with_retry(file_path, max_retries=5):
    for attempt in range(max_retries):
        try:
            with open(file_path, "rb") as f:
                data = f.read()

            base64_string = base64.b64encode(data).decode("utf-8")

            response = client.responses.create(
                model="gpt-4.1-mini",
                input=[
                    {
                        "role": "user",
                        "content": [
                            {
                                "type": "input_file",
                                "filename": os.path.basename(file_path),
                                "file_data": f"data:application/pdf;base64,{base64_string}",
                            },
                            {
                                "type": "input_text",
                                "text": f"What is document about? (Attempt {attempt + 1})",
                            },
                        ],
                    },
                ],
            )
            return response
        except Exception as e:
            print(f"Attempt {attempt + 1} failed: {str(e)}")
            if attempt == max_retries - 1:
                raise

# Usage
file_path = "example.pdf"
try:
    result = upload_with_retry(file_path)
    print("Upload successful:", result)
except Exception as e:
    print("All attempts failed:", str(e))

By implementing these suggestions and following best practices, you should be able to improve the reliability of your file uploads to the Azure OpenAI API. If issues persist, please consider reaching out.

Thanks

Hi Ravada Shivaprasad

I ran that and all attempt failed.
the pdf is very small (less than 1MB) and if I just remove these lines

{
    "type": "input_file",
    "filename": os.path.basename(file_path),
    "file_data": f"data:application/pdf;base64,{base64_string}",
},
      

it works so it is not network or server side disruptions and the problem is just with the file upload

Also pdf is https://arxiv.org/pdf/2306.13230
Did you try the code you wrote and does it work on your end?

Attempt 1 failed: Error code: 500 - {'error': {'message': 'The server had an error processing your request. Sorry about that! You can retry your request, or contact us through our help center at [email protected] if you keep seeing this error. (Please include the request ID 0a9d8187-0b31-4d43-9d7d-46848431cf77 in your email.)', 'type': 'server_error', 'param': None, 'code': None}}
Attempt 2 failed: Error code: 500 - {'error': {'message': 'The server had an error processing your request. Sorry about that! You can retry your request, or contact us through our help center at [email protected] if you keep seeing this error. (Please include the request ID 8cbdbb3b-35e8-4085-8863-864c3ec7fc31 in your email.)', 'type': 'server_error', 'param': None, 'code': None}}
Attempt 3 failed: Error code: 500 - {'error': {'message': 'The server had an error processing your request. Sorry about that! You can retry your request, or contact us through our help center at [email protected] if you keep seeing this error. (Please include the request ID d0b20d73-05ce-4a9f-95c6-8db3ab493238 in your email.)', 'type': 'server_error', 'param': None, 'code': None}}
Attempt 4 failed: Error code: 500 - {'error': {'message': 'The server had an error processing your request. Sorry about that! You can retry your request, or contact us through our help center at [email protected] if you keep seeing this error. (Please include the request ID 45841f53-8b6e-46be-bb33-32f49ace9dac in your email.)', 'type': 'server_error', 'param': None, 'code': None}}
Attempt 5 failed: Error code: 500 - {'error': {'message': 'The server had an error processing your request. Sorry about that! You can retry your request, or contact us through our help center at [email protected] if you keep seeing this error. (Please include the request ID a7b4b08a-4fef-40ef-85a0-99b7b327c7bd in your email.)', 'type': 'server_error', 'param': None, 'code': None}}
All attempts failed: Error code: 500 - {'error': {'message': 'The server had an error processing your request. Sorry about that! You can retry your request, or contact us through our help center at [email protected] if you keep seeing this error. (Please include the request ID a7b4b08a-4fef-40ef-85a0-99b7b327c7bd in your email.)', 'type': 'server_error', 'param': None, 'code': None}}

I tried Ravadas Solution too, and it does not work. 'The server had an error processing your request. Sorry about that! You can retry your request, or contact us through our help center at [email protected] if you keep seeing this error.

In addition I used Postman to send exactly the same request and I got another error:

"You must be logged in to use this feature."

I find it quite amusing that the API includes internal support email addresses in error messages...