PDF file as text not supported?

[etienneroger]

Trying to send a pdf as base 64 text but getting this error. Is pdf not yet working or is there something not right in my request?

============
{
“error”: {
“message”: “Invalid Value: ‘file’. This model does not support file content types.”,
“type”: “invalid_request_error”,
“param”: “messages[0].content[1].type”,
“code”: “invalid_value”
}
}

Message sent to gpt-4.1-2025-04-14 using the completion API:

{
“max_completion_tokens”: 16384,
“messages”: [
{
“content”: [
{
“text”: “This is the content of document Austenic.pdf”,
“type”: “text”
},
{
“file”: {
“file_data”: “data:application/pdf;base64,snip”,
“filename”: “Austenic.pdf”
},
“type”: “file”
},
{
“text”: “describe”,
“type”: “text”
}
],
“role”: “user”
}
],
“model”: “gpt-4.1-2025-04-14”,
“reasoning_effort”: “high”,
“store”: false,
“stream”: false,
“temperature”: 1
}

[_j]

Did you have any intention of actually adding the PDF file contents with an f-string?..

"file_data": f"data:application/pdf;base64,{pdf_files_list_b64[0]}"

from openai import OpenAI
client = OpenAI()

response = client.chat.completions.create(
  model="gpt-4.1-2025-04-14",
  messages=[
    {
      "role": "system",
      "content": [
        {
          "type": "text",
          "text": "A helpful and honest assistant"
        }
      ]
    },
    {
      "role": "user",
      "content": [
        {
          "type": "text",
          "text": "What are the price ratios between gpt-4.1 AI model sizes?"
        },
        {
          "type": "file",
          "file": {
            "file_data": "data:application/pdf;base64,JVBERi0xLjMKJcfsj6IKNSAwIG9iago8PC9MZW5ndGggNiAwIFIvRmlsdGVyIC9GbGF0ZURlY29kZT4+CnN0cmVhb///////wMDAgbiAKMDAwMDAwMDQzOCAwMDAwMCBuIAowMDAwMDAwNTM4IDAwMDAwIG4gCjAwMDAwMDA0NzkgMDAwMDAgbiAKMDAwMDAwMDUwOCAwMDAwMCBuIAp0cmFpbGVyCjw8IC9TaXplIDExIC9Sb290IDEgMCBSIC9JbmZvIDIgMCBSCi9JRCBbPDlERThEOTdEODhBQzZGNEZDNTgyNUQ4RjM4M0M0NzBCPjw5REU4RDk3RDg4QUM2RjRGQzU4MjVEOEYzODNDNDcwQj5dCj4+CnN0YXJ0eHJlZgoxMjc1NzAyCiUlRU9GCg==",
            "filename": "test2.pdf"
          }
        }
      ]
    },
  ],
  max_completion_tokens=2048,
  top_p=0.5,
)

reasoning_effort is also not a supported parameter for that model.

update openAI library if using.

[etienneroger]

ticking off reasonning fixed it! Thanks,

[M_Afifudin_Abdullah]

im literally doing this but somehow still not working

rc/consumers/application-document-consumer.ts:204:9) {
  status: 400,
  headers: {
    'apim-request-id': '6efeb34e-7686-4577-af70-b3b843466936',
    'content-length': '218',
    'content-type': 'application/json',
    date: 'Mon, 16 Jun 2025 05:09:50 GMT',
    'strict-transport-security': 'max-age=31536000; includeSubDomains; preload',
    'x-content-type-options': 'nosniff',
    'x-ms-deployment-name': 'gpt-4o',
    'x-ms-rai-invoked': 'true',
    'x-ms-region': 'Japan East',
    'x-ratelimit-limit-requests': '4500',
    'x-ratelimit-limit-tokens': '450000',
    'x-ratelimit-remaining-requests': '4499',
    'x-ratelimit-remaining-tokens': '448927',
    'x-request-id': 'a6f7d2fc-005b-4289-baad-1edd928612e5'
  },
  request_id: 'a6f7d2fc-005b-4289-baad-1edd928612e5',
  error: {
    message: "Invalid Value: 'file'. This model does not support file content types.",
    type: 'invalid_request_error',
    param: 'messages[0].content[1].type',
    code: 'invalid_value'
  },
  code: 'invalid_value',
  param: 'messages[0].content[1].type',
  type: 'invalid_request_error'
}

using it through azure

[_j]

On Azure you will need to specify the latest preview API version available in order to get API parameters that keep pace with OpenAI.

[lococroco]

Did you get it running? I just tried it with API version 2025-04-01-preview on azure and got the same error as M_Afifudin_Abdullah. Which API version did you use?

[Gabriel8]

I don’t think azure supports pdf uploads from what I’ve seen, even on 2025-04-01-preview , it doesn’t work and there is no evidence of it being supported. Ignore @_j

[thiagopo]

Same here. I tried to use gpt-4o deployment on Azure OpenAI API version 2025-04-01-preview and got same error.

Even with API V1 Preview ( <resource>.openai.azure.com/openai/v1/chat/completions?api-version=preview, passing "model": "gpt-4o" on body) the error was "Invalid Value: 'file'. This model does not support file content types."

[_j]

@Gabriel8

there is no evidence of it being supported. Ignore @_j

Evidence of it being supported:

When you send the query string for the version or simply “preview”.

The key is that this is only validated through on Responses API. Chat Completions, you get audio input instead:

      "chatCompletionRequestUserMessageContentPart": {
        "oneOf": [
          {
            "$ref": "#/components/schemas/chatCompletionRequestMessageContentPartText"
          },
          {
            "$ref": "#/components/schemas/chatCompletionRequestMessageContentPartImage"
          },
          {
            "$ref": "#/components/schemas/chatCompletionRequestMessageContentPartAudio"
          }
        ]