Regarding the issue of calling the Gpt_4_vision_preview model

[yolo-lxf]

When I use the Models.Gpt_4_vision_preview model, the responses I get are always strange. For example, when I ask "Are you a vision model?", the reply I get is "No, I am not, I am a large language model." I suspect that although I specified Models.Gpt_4_vision_preview, it is not the model that is ultimately being called.

[belaszalontai]

Have you tried this?

I think Vision model is introduced for analyzing images not chat. The response is definitely coming from the requested vision model even if we don't like the answer.

[belaszalontai]

And if you provide image to this model then even the question was not related the answer looks promising :)

[yolo-lxf]

Yes, I have tried it, but the photos I pass through the method never correspond to the analysis results it gives. Could it be that there is an error in the way I pass the photos?
This is my code block:

[belaszalontai]

Vision API haven't implemented yet @ betalgo/openai SDK. You have to wait for it..

[yolo-lxf]

Thank you so much for responding to my issue. I really appreciate your help and the work you're doing on this open-source project.

[belaszalontai]

Why your code does not work? Because in the currently implemented chat completion request handles the content property as a text (string) and you want to handle it as a json array. In the request to openAI , your content will not be a json array but a simple string contains a lot of weird characters.

[yolo-lxf]

Oh, I see. Then I'll wait for the API update.