Comment by ibejoeb

Ah, you're probably one to ask. They say "queries on some topics will instead receive a response from our next-most-capable model, Claude Opus 4.8." Are they transparent about when that happens, and is it priced at the rate of the underlying model?