Alright, I finally gave Fable a spin today

I am a 10 year experienced cloud architect with a DevOps background. I finally decided to give Fable a try during my on-going production soft-launch of my project. Been working on it for the past year, starting off with Gemini 2.5 Pro, Claude 4 fam and so on. I've learned a heck ton. The point of the project wasn't so much a business idea, but as a means to learn how to best use LLMs while covering a personal tool gap (AI bank statement analyst). Kinda of like a knowledge-as-a-product project.

Well, I have found my rhythm and pacing with Opus/Sonnet over the months. I am more of an acting technical project manager. I am writing up the epics, features and user stories. That's what I feed into my daily sessions. 1 session = 1 user story. This roughly equates to using around 15%~ of my 5-hour usage limit (Opus 1M, 5x plan) just for planning, 5%~ execution and some more for whatever follow ups. It all comes down to managing work (which is why I understand some developers out there missing out on the actual coding). I finish my session always under 40% 5-hour usage limit.

My fable session just used around 35% of my usage limits just for planning! I know it's more expensive, and I know it's more token hungry (feels like it reads everything beyond what I scope it to), but it's still eyebrow-raising to see. I did not need to use Fable for my task, I was just trying it out.

With that said, it's a very "clever girl". It caught bugs along the way. It's thought process feels thorough, goes deeper and broader than Opus does (token tax comes into play here). I and, perhaps, many others out there will need to identify the prompts/tasks that require Fable. It's just dang hard because it's not obvious when certain actions are done by Fable vs. Opus. It's going to take time.

Anyone out there with a similar experience with Fable?