Post

I used it during the week to review a large risky piece of work across multiple PRs. It was _really_ good, like pairing with a Staff developer. It’s not an everyday model though: it’s so tenacious that it happily ran for about 1.5hrs. Gonna miss it for the big risky stuff

I imagine they’ll work with the government to figure out a version of it they can ship How would you compare it to working with Opus? a bit of a big question, I realize

I think a more concrete difference: Opus would say "the most likely root cause is" because otherwise it would spend a long time correlating data but Fable would have chased down all the production data to prove whether it was or wasn't. It just didn't really care about how long it spent on a task

tbh, I've rarely noticed any differences in Opus model so maybe I'm not the most sensitive person. But I did think Fable was more concise, more precise, and there just didn't appear to be a complexity horizon where it would give up or gloss over things.

For me personally, Fable didn’t live up to the hype. It just eats up all my tokens 😅

ha maybe it’s good I didn’t then! I was into overage two days this week

missed my chance I guess 😂