Codex used 18M Input tokens in a day
Resolved 💬 0 comments Opened Apr 20, 2025 by saleweaver Closed Apr 20, 2025
I've let codex analyze a midsize project, and since it failed to write results, I may have restarted and asked for the same a couple of times. This resulted in producing 18M input tokens, for ingesting a html file of 300 lines, not too much text actually.
Mainly cached input using the 4.1 model. Is this normal?