Pass PDF to analyse images in OpenAI (or other formats and other LLMs)

Hi

I have a problem:
When I had one-page PDF files, I could easily convert them to JPG and have them analysed by OpenAI.
My problem comes when I get PDF with more than 1. How to convert and analyse them by an LLM such as OpenAI?

I can accept other alternatives: convert to binary, text, OCR, etc. I can also accept other AIs, such as Gemini, Deepseek, etc.

What are these documents? They are varied: Air tickets, invoices, purchase documents, etc, with different formats and different data. I need to read each document and define what they are. OpenAI had some trouble reading some information, even when they were only invoices with same structure.

Do you want to analyze the content in your PDF? You can extract text from your PDF using the following tools and then provide the extracted text to ChatGPT for analysis. (Each tool supports up to 1,000 file conversions per month.)

1 Like

Hi

I got a solution (I think):
Instead of converting to image and send to OpenAI Analyze image, I download the image, use OpenAI Upload file, Add file to Vector Store and then use an Assistant.

It is a nice solution as the file remains unconverted, spend less credits on other sources, but may add a couple more operations to the path. I’ve also tested and it’s better at reading text than using vision, that is, I had less mistakes.

2 Likes