Fable, Opus, Sonnet, Haiku: how do you decide which Claude model to use?
retained text
Right now the lineup is Fable 5.1, Opus 5.5, Sonnet 5.5 and Haiku 4.5, and I keep going back and forth on what to use for what. I run my Hermes agent on Haiku because it’s cheap and quick. What I don’t know is whether a bigger model would be worth the extra cost for that kind of work, or whether I’d just be paying more for the same results. What are your actual rules? By task type, by cost, by speed, or do you just pick the best one and move on? And has anyone found a case where the smaller model beat the bigger one?
document fetched 2026-10-14 00:22 · canonical https://reddit.com/r/ClaudeAI/comments/1x4q63a/fable_opus_sonnet_haiku_how_do_you_decide_which