I'm glad Jake's run the numbers — but do make sure you validate your own use-case rather than just following them. My personal experience has been somewhat different, but also my use-case appears to be pretty different so that's OK.

That said, I last benchmarked over a year ago, maybe I should re-evaluate. My experience was that AVIF was too slow for on-the-fly encoding, while JXL was absolutely fine and JPEGLI was also pretty useful.