So, I built tokount. It is almost entirely built by prompting the replit agent. Except for the tokenizer, which was written with some help from chatGPT o1-preview.
It cost me approx. $45 just for prompting the replit agent, and a few additional dollars for database and VM. Took about 2 full days and did it on a mix of phone and laptop. Needed the laptop for tokenizer, most other features could be built just on the phone.
Conclusion: I’m definitely better off using the API. It’s surprising how much text is needed to reach 1M Tokens.
If the title includes the name of the site, please take it out, because the site name will be displayed after the link.
Otherwise please use the original title, unless it is misleading or linkbait; don't editorialize https://news.ycombinator.com/newsguidelines.html