This is an interesting comment because it makes me want to ask the question, what is the use for fable then? For me, GPT 5.4 is enough when using recursive-mode. I do appreciate GPT 5.5 Pro for some larger research, architecture, planning tasks though. I think that's what Fable is for. A very small % of total work.
>planning tasks though. I think that's what Fable is for.
this is how i've been using it, and where i've found it really excels over anything else i've tried. get fable to write a plan, and get something cheaper to follow the plan. the code fable writes isn't significantly better than the code opus writes, as long as they're both following the same plan. but a plan written by fable is much better.
What is the use of a genius who does great work but doesn’t document their results, only produces them? How much do companies enjoy having a structural dependency on someone like that?
I think it depends on your workflow. I've had a great experience with the trial. I work in research, and have set up something similar to Kaparthy's auto research. I, with Fable, have managed to get an image generation model down from 80M parameters to 10M and keep the quality of the generated images on par (similar FID). And, importantly, every change was modular, explained by Fable, reviewed by myself, and understood and documented. If not understood, I read relevant docs until I did it didn't accept it as part of the plan. So it ended up being a simple composition of existing ideas which I had previously encountered, but stacked much more rapidly than I could have.
The structure of the code is easily readable as I enforce concenventions followed by good libraries. And I can easily plug in new datasets. It's pretty good frankly.
That seems valid in today's world. Right now it's expensive, slow, and accurate. I imagine in the fairly near future it will be cheap, slow, and accurate, and that'll be a great opportunity to let it run on anything time-insensitive.
Re current use-cases: in addition to planning, there's also some tasks which Opus just can't complete but Fable can. Multiple times I've spent hours in combination w/ Opus trying to debug some particularly nasty nondeterministic issue, only to have Fable nail it in 20mins while I walk the dog.
Definitely not just a classifier layered on top, although there is one of those as well. Pretty sure it's different post-training / finetune run and the model weights are different between them. Which is bound to have effects on the model output even in general use, although I'd imagine they have pretty stringent evals to make sure it doesn't regress too much in things they don't want to restrict.
