A Reddit post claiming 75% output token savings sparked 400 comments and multiple GitHub repos dedicated to making AI grunt its way to efficiency.

Read Full Story at Decrypt.co →