Fable is mega nerfed?
2 chats on different projects, each routed 4-10x more tokens through Opus than Fable. Fable did only 20% of the work...
wtf. disappointing. Is anyone else seeing this?
Fable is mega nerfed?
2 chats on different projects, each routed 4-10x more tokens through Opus than Fable. Fable did only 20% of the work...
wtf. disappointing. Is anyone else seeing this?
Some users defend Fable as unchanged and well-suited to subagent tasks while others call the model nerfed or a scam after reports it handles only 20% as much work as Opus.
No Digg Deeper questions have been answered for this story yet.
This is for a notes app and RTS game btw.
No idea why this would trip the safety filters.
Fable is mega nerfed?
2 chats on different projects, each routed 4-10x more tokens through Opus than Fable. Fable did only 20% of the work...
wtf. disappointing. Is anyone else seeing this?

@NickADobos $600 and it was literally all opus 4.8... such a scam. Can't wait for 5.6 to fuck this fable scam shit off.

@NickADobos Is it spinning up sub-agents? It might choose other models for those.

@NickADobos Probably Opus subagents, right? Fable is supposed to be subagent happy. Which is exactly what we want. You don't want Fable renaming a variable in a file any more than you need a human doing it.

@NickADobos you need to specify your /agents workflow to fable. I can share mine if you want

@NickADobos it sucks

@NickADobos i didnt get to use it first time -- and this is an LLM project(fucking about with qwen 27b)

@NickADobos Yes it consumes more tokens, but nerfed? Not for me, I use it and it's already exactly like the previous version

@NickADobos 😶

@NickADobos i was switched to opus mid task too 😓

@NickADobos Same. 20% tracks with my sessions. When did Fable start routing this conservatively?

@NickADobos How do you get it to spit out the true metrics? I am using Claude on mac desktop app