To be fair the new models are likely 2x as expensive as before
Supposedly much less expensive per task (more token efficient and cheaper)
Supposedly much less expensive per task (more token efficient and cheaper)