Maybe try the new version 4 with vision? Seems like you’d avoid all the weird formatting issues if you just upload an image of the pdf.
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| Api generating wrong responses and skipping data I want | 1 | 620 | April 8, 2024 | |
| Need to come of finish_reason: length Max token alternatives | 5 | 1892 | April 10, 2024 | |
| GPT-4.1 extracts table details from PDF | 4 | 315 | August 6, 2025 | |
| Finance Agent -> reading Pdf scans (failing) | 0 | 137 | March 12, 2025 | |
| Gpt-3.5-turbo , to generate correct responses | 8 | 1747 | January 2, 2024 |