Help in API PDF Extract information

Hello everyone, I’m building an API to extract data from an energy/water bill.

It works by inputting a .PDF file and outputting JSON. Since I’m using the wizard, I’m using the GPT4o, TOP P: 1.0 and Temperature 1.0 models.

The problem is that sometimes the data extraction is missing some data. When I attach the same .pdf file directly to chatgpt, the extraction is performed correctly.

I would like some help on which model is best for this type of task.

Thank you.

Note: I used Google Translate to write the text, sorry for any typos;

You may want to reduce the temperature a little since you only want to extract information and you must specify well with the prompt what you want to extract specifically. I did a similar project a few months ago.