On Monday morning, I almost made the most expensive model my default

Last Monday, I was preparing my weekly personal IP plan in Hangzhou. Client details, topic ideas, and the content schedule were all squeezed into one spreadsheet. My instinct was to hand everything to the strongest model, but I opened the pricing page and backed out: complex tasks are worth upgrading for, but everyday organization does not need to burn money on every item. I got stuck in this “more expensive means less work” phase too. Later, I learned to sort the task first, then choose the model.

Not two new buttons, but two kinds of work partner

Both Fable 5.2 and Opus 5.5 handle text, code, and automation tasks. Put simply, Opus is more like an on-demand execution partner, while Fable fits hard tasks that need repeated reasoning, review, and sustained progress. In the original article, author Peter Yang shows two typical examples: he used Opus to build an “inbox and schedule manager,” and Fable to design and develop a playable 3D kart game. The stronger the model, the higher the cost per use; the article says Fable costs about twice as much as Opus when billed by usage.

Try this today: spend $0 on a 30-minute task split

Money: use the Claude Code account you already have and skip the upgrade while testing. The $600+ allowance mentioned in the article is a subscription benefit, not a requirement for trying this. Time: about 30 minutes. Technical barrier: being able to copy a prompt that asks the model to break down a task; no coding knowledge needed. First step: log in to Claude Code, create a blank project, type “split this task into daily execution and complex problem-solving,” then paste in the most mentally demanding thing on your plate.

My example split looks like this: meeting notes, client copy edits, and small website features go to Opus. Comparing approaches, working through long source materials, and continuously revising a complete product move to Fable. Instead of copying someone else’s “strongest model list,” I start with real tasks. If a tool is not something everyone needs, there is nothing wrong with not upgrading yet.

Choose by stage; there is no need to get there in one step

If I am just getting started: I would use Opus for daily tasks and save my energy for finding clients. Once a task remains unclear after 30 minutes of work, I would try Fable.
If I already have 1-2 clients: I would turn quotes, revisions, and delivery checklists into three fixed task types. Opus handles the routine; Fable steps in for strategy or one-off difficult problems.
If I am scaling: I would track each model choice, usage time, and number of manual rework rounds. After two weeks, I would use the data to decide which task types deserve a more expensive model instead of guessing.
If I do not want the hassle: I would stick with one default model for a week and switch only when I actually hit a bottleneck.