"Cannot specify both model and engine" in app-request

[Stve_O]

Hi, I´m trying to make a small iOS app for myself in flutter.
I have a paid subscription and also a valid API key. In my app I want to let describe the Ai a photo. Its working well in the chatGPT app…so it should also work with the API right?
I´m using the endpoint url: httxx://api.openai.com/v1/chat/completions
and as model: gpt-4-turbo

In my app there´s also showing up the error: “Cannot specify both model and engine” and I dont know why.
Here´s the sentence in the flutter file…maybe someone can help me out?

`Future<String> _getImageRecognition(String imagePath) async {
    // 
    final url = Uri.parse('https://api.openai.com/v1/chat/completions');

    // 
    final bytes = await File(imagePath).readAsBytes();

    // 
    final base64Image = base64Encode(bytes);

    // 
    try {
      final response = await http.post(
        url,
        headers: {
          'Authorization': 'Bearer $apiKey',
          'Content-Type': 'application/json',
        },
        body: jsonEncode({
          'model': 'gpt-4-turbo', // 
          'messages': [
            {
              'role': 'system',
              'content':
                  'XXXXX'
            },
            {
              'role': 'user',
              'content':
                  'XXXXX: $base64Image. '
            }
          ],
          'max_tokens': 100 // Beispielparameter; 
        }),`

[PaulBellow]

Heya. Welcome to the forum.

I believe you’re not sending the image correctly. Check the API docs… Here’s a good post with some hints…

How to load a local image to gpt4 -vision using API

In response to this post, I spent a good amount of time coming up with the uber-example of using the gpt-4-vision model to send local files. Stuff that doesn’t work in vision, so stripped: functions tools logprobs logit_bias Demonstrated: Local files: you store and send instead of relying on OpenAI fetch; creating user message with base64 from files, upsampling and resizing, for multiple images per message; calling with streaming in “raw” mode; parsing, loading headers and displaying chunk…

[Stve_O]

Thanks! Might be a solution…pretty hard for a complete noob in this field

[Stve_O]

Ok, I made it so far. But now the token request is with 473018 way too big than the max of 30000.
I tried to reduce the image size (25%, 128x128px) but nothing works. The requested tokens are always the same.
But it should be possible to upload a photo from the smartphone to analyse it, or? Can´t believe that this is so token-wasting.

[_j]

Here’s a quicker guide of what you must send to chat completions:

400 Error when I include base64 image in a prompt

You don’t modify the contents of a base64 unless you want to corrupt it. This is the format of message that must be sent to chat completions: [ { "role": "system", "content": [ {"type": "text", "text": system_prompt} ] }, { "role": "user", "content": [ {"type": "text", "text": "Describe this art style."}, { "type": "image_url", "image_url": {"url": f"data:image/jpeg;base64,{e…

[Stve_O]

Thanks a lot! That was a good hint…now it works!