Claude Fable 5.2 is a massive downgrade and truely a massive dissapointment and feels unusable and VERY unpredictable.
retained text
Ever since Claude Fable 5.2 came out, I've been trying to build my project with Fable as the orchestrator, project manager that checks the work of agents, sends agents new tasks, and has the whole plan of the project in mind and the big picture. But it is just not even as good as Claude Opus 5.5. (I use both models with reasoning set to high) Fable It is taking forever to just make a simple little thing/a simple feature of the thing its supposed to build and . It's not able to make anything really fast. It takes weeks for minimal things and it constantly keeps rechecking every single little change and runs tests all the time for every single little thing. I already told it to do less but it keeps just WASTING on unnecessary little details which it keeps rebuillding or changing stuff and keeps finding errors and WAISTS hundereds of tests and tasks on unnecessary and VERY small details and tests and keeps coming back and to the same things and does not even fix the isse entierly just a little bit ... and doing even more tests after every little change. It already burned through the free weekly, the Fable reset, and the usual Ultralimits without really making any meaningful progress because it keeps testing again and again and keeps waisting limits without progress even though is does A LOT... but finishing goals on the Plan we made beforehand is not one of the things it's doing... It's just such an incompetent model. Even Opus 5.5 was way better and did tasks or did work on long running Taks for day plans without any issues. Also Fable 5.2 is so insanely bad at expressing itself. It just keeps writing very incoherent answers and it's just extremely hard to understand what it's trying to tell you. Even if you tell it to explain the thing to you, it just explains it like shit and what it explains is very incoherent. It also thinks that I know all its context and can also know all things that it knows too. It just asks about things without context, literally out of context, or just drops random stuff in questions and is just very incoherent. It's just extremely bad at expressing itself. It is very bad at following instructions the way Opus 5.5 did, at least when you gave Opus 5.5 a task or something you wanted it to build. Opus 5.5 was very good at understanding and also building the thing you wanted and just filling in the blank spaces. Fable feels like it's capable of doing things but it is just so insanely weird to prompt it. It mistakes examples, which are clearly marked as examples, and is even told that those are just examples. It just keeps misunderstanding and uses examples as if I wanted exactly this thing. It just takes everything way too literally and is so bad at understanding what the user wants from it. Also if you give it rules, it often just follows them to such an extreme point that it's just hindering itself. It's just bad overall. I think that's a pretty long rant about Fable 5.2 but I expected very much of this model and was very happy when it was first released. After weeks of testing it just proved that it is just a downgrade from the previous models. Yeah it's faster but I really don't like it and hope they will change something in the future. It's unusable this way.
document fetched 2026-11-18 00:22 · canonical https://reddit.com/r/ClaudeAI/comments/1xyuqtw/claude_fable_52_is_a_massive_downgrade_and_truely