Issue with Uploading Arabic PDFs to File_Search Tool in Assistant API

Hi everyone,

I’m developing a personal assistant using the OpenAI Assistant API, and I’ve integrated the file_search tool to read and search through attached PDFs. I’m working with a combination of fully English and fully Arabic PDFs. The English PDFs are handled without any issues, but I encounter an error when trying to attach or process the Arabic PDFs.

Here’s what I’ve set up:

I’m using a vector store to store embeddings from both English and Arabic PDFs.

The file_search tool works well with English PDFs, but when uploading Arabic ones, the assistant either returns an error or fails to read the content.

I would appreciate any insights on how to correctly handle Arabic PDFs or if there are specific configurations I should look into for multilingual support.

Thanks in advance for the help!