Opus 4.5 seems to think a lot less than other models, so it’s probably not as many tokens as you might think. This would be a disaster for models like GPT-5 hig

Opus 4.5 seems to think a lot less than other models, so it’s probably not as many tokens as you might think. This would be a disaster for models like GPT-5 high, but for Opus they can probably get away with it.