@nicbstme I tend to think we’ll build a general model that can build specialist models/harnesses/tools that outperform it in given tasks in some regard (cost/performance) and then we will rl on that and have even better general models. Which is kind of what’s happened with coding.
If Fable isn't in the Claude Code sub, what's the point of the sub for developers? When a new frontier tier drops, work moves up to it. Everything used to happen at Sonnet level; now almost all of it happens at Opus. Will subs even include future smaller models?
So if cowork is 2x usage limits until June 9th and fable uses 2x limits… then fable is the same usage as opus was in cowork?
Or am I missing something?