I’m trying to upload a bunch of text to my custom GPT. I have a master file that I split up into 20 .txt files (20 is max number you can have, right?). Each part is about 8.5 MB. I get a red popup that says “Unable to upload file.txt”.
2 Likes
The maximum content that can be included is two million tokens per file (only documented on API’s Assistants retrieval).
For plain English text, the most efficient, you are about 4 characters per token. Eastern languages and code will have a reduced compression ratio, limiting you closer to 5MB files or less.
That is a bit larger than the amount of text that online token counters can number-crunch if you really want to approach the limit. You can use the Python tiktoken library to measure the token count of BPE-encoded files.
1 Like