How to use ChatGPT for OCR
- Tap the + icon in the prompt area and choose Add photos and files, or drag the image into the chat[1].
- Ask for what you want, for example the full text of the page, line by line.
- Copy the answer from the chat.
ChatGPT accepts PNG, JPEG and non-animated GIF images[2]. It also reads text and images inside PDFs, including charts and diagrams[2].
What limits does OpenAI list?
- Image size
- 20 MB per image[1]
- Documents
- Up to 512 MB per file, capped at 2 million tokens[2]
- Upload rate
- Up to 80 files every 3 hours. Free users get 3 file uploads a day[2]
- Images per chat
- Depends on image size and the text that goes with them[1]
OpenAI's own FAQ names the weak spots for image input. The model does not perform as well on text in non-Latin alphabets such as Japanese or Korean, may misinterpret rotated or upside-down text, and may generate incorrect descriptions in some cases[1]. Images are resized before analysis[1].
Can ChatGPT read handwriting?
ChatGPT reads handwriting in a photo the same way it reads print, as part of the image. OpenAI's FAQ says that if an image is unclear, the model does its best to interpret it and the results may be less accurate[1]. A clear, flat photo of each page, shot straight on, gives it the best start.
LoreAI is built for this job. It reads cursive and printed handwriting from letters, diaries and class notes and returns the text in a Word file, page by page in the original order. For more on old letters, see handwriting to text.
ChatGPT and LoreAI side by side
| What you need | ChatGPT | LoreAI |
|---|---|---|
| File types | PNG, JPEG, GIF images, plus PDF, DOCX and TXT documents[2] | PDF, JPG, PNG and HEIC |
| Size | 20 MB per image, 512 MB per document[2] | Free 3 MB, Basic 10 MB, Pro 25 MB |
| Length | Up to 2 million tokens per document[2] | Free 3, Basic 30, Pro 100 pages in one file |
| Scanned pages | Read as part of the PDF[2] | Every page read with OCR, in order |
| Photos | Attached in the chat[1] | Uploaded directly, HEIC included |
| What you get | An answer in the chat | An editable Word file (.docx) with headings, lists and tables |
| Phone | Web and supported mobile apps[2] | Any phone browser |
| Price | Free and paid plans, with plan limits[2] | First document free, then Basic at $12.99 a month |
When to use ChatGPT, and when a document OCR
- Use ChatGPT to ask questions about a photo: what a sign says, what a receipt totals, what a paragraph means.
- Use ChatGPT when you want a summary of a scan, not the scan itself.
- Use LoreAI when you need the document back as a document: every line, the tables as tables, the headings in place, ready to edit in Word.
- Use LoreAI when the page also needs translating, into one of 105 languages in the same step.
For translating whole files with ChatGPT, see ChatGPT for document translation. To compare handwriting tools, read the best handwriting OCR comparison.
Try OCR to Word on your own page
For one page, the free image to text tool gives you the full text and a Word download in a few seconds, with no account. For a whole document, upload it to the LoreAI app. Your first document is free, up to 3 pages.
Questions people ask
Can ChatGPT read text from an image?
Can ChatGPT transcribe handwritten notes?
It reads handwriting as part of the image. OpenAI notes that unclear images may give less accurate results[1]. LoreAI reads cursive and printed handwriting into a Word file.
Is ChatGPT OCR free?
Image and document uploads are available on the Free plan, which allows 3 file uploads a day[2].
Is ChatGPT good at OCR for Japanese or Korean?
OpenAI says the model does not perform as well on text in non-Latin alphabets such as Japanese or Korean[1]. See Japanese OCR and Korean OCR.
Sources
- OpenAI Help Center: ChatGPT Image Inputs FAQ, accessed November 6, 2026
- OpenAI Help Center: Uploading files and audio to ChatGPT, accessed November 6, 2026
By Matt Mountain ยท