Sometime in the next thirty-six hours, Fable 5 will be removed from Claude subscription plans. If you have been using it inside your regular Claude plan, running the hard jobs and watching it hold a single thread across a long piece of work, that arrangement ends tomorrow. The change is narrower than the notice makes it sound. Fable stays available: it moves to the API, where access is metered, and you pay for what you use, token by token.

I want to put real numbers on that. On the API, Fable charges $10 per million tokens of input and $50 per million tokens of output. In practice, a single long research run or a heavy document job might cost you a few dollars, sometimes into the low tens if you let it work all afternoon at full effort. The cost is small enough to absorb and large enough to make you ask whether a given task is worth it, a question the flat monthly fee lets you skip. So the thirty-six hours ahead of you are the last stretch, where extravagance is already paid for.

Here is how I would spend them, and the logic holds no matter which way you look at it. The most valuable thing Fable can do before the meter starts is help you build things that outlast the subscription. Use the expensive brain while it is included to make the assets you will run on cheaper models tomorrow.

Start with Skills, since they are the one thing you make today that keeps paying out. A Skill is the packaged, reusable form of a workflow you keep rebuilding by hand, the “final form” of a prompt you have finally stopped fighting with. You hand Claude a Skill once, it loads automatically when the task shows up, and the weekly re-explanation of your proposal format retires. The move today is to point Fable at how you actually work. Something like: look at my recent projects and the prompts I keep rewriting, find the five workflows I rebuild from scratch every week, and turn each one into a Skill with clear instructions and a defined output. One good Fable session can spot the patterns you stopped noticing and draft the Skills for you, and those Skills keep working afterward on Sonnet or Opus.

With a Skill or two banked, turn Fable loose on the one hard job you have been avoiding. It is at its best on long, ambiguous, multi-step work that a smaller model fumbles: a full competitive audit, a pile of scattered research turned into one coherent read, a brand reconciliation across six documents that contradict each other, a first real draft of the proposal you have been dreading. Those are exactly the jobs that cost the most per token once billing kicks in, so run the one or two that matter most now, while the meter is off.

The real upside for a creative leader is orchestration. Fable is strong enough now to sit in the producer’s seat. It holds the whole project in its head, hands the scoped pieces to cheaper models working in parallel, and checks their output before it fits everything back together. Fable is the one built to keep the whole picture. Point it at a campaign you are about to launch and let it break the work into research, copy, and asset briefs, send those out, and assemble what comes back. You stay the creative director, while it runs the room.

It also earns its keep on the numbers, which is where I would send any owner with an hour to spare tonight. Fable is the strongest model Anthropic has tested on finance work, and the use that carries over to every business is a spreadsheet audit. Open the workbook behind your budget, your last invoice batch, or the pricing model you built at 2AM and have not trusted since, and ask Fable to check every formula and tell you where the math stops making sense. It reads the tables buried in the file and catches broken references before they reach a client. The point is a second set of eyes that does not get tired on row 400.

(One caution before you feed it anything sensitive. In the API, Fable holds your data for 30 days, so keep confidential client financials out of it until you have cleared them against your own rules.)

Whatever you point it at, a few habits change the output more than any clever phrasing. It starts with the why. Fable gets noticeably sharper when it understands intent, so tell it who the work is for and what a win looks like before you tell it what to do. “I am building a rebrand pitch for a skincare founder who is skeptical of agencies. She needs to feel we understand her category better than she does, and, with that in mind, draft the competitive landscape.” That framing gives the model something to aim at. Without it, Fable guesses at your intent and fills the gaps with something generic.

Once it knows the why, step out of its way. Hand it a destination rather than a list of steps you want executed in order, since Fable plans well on its own. Stay involved while it is still planning, and leave it alone once the build is underway. Before you let it run unsupervised, though, make it hunt for your blind spots. Ask it to interview you one question at a time, starting with the answers that would change the whole direction. That step surfaces the assumptions you did not know you were making, before they show up as vague output an hour later.

It also helps to draw boundaries. At full effort, Fable will happily add sections you never asked for or tidy work you wanted left alone, so one line keeps it in check: when I am thinking out loud rather than asking for a deliverable, give me your read and then stop. Two smaller notes protect both your money and your output. Let the reasoning stay behind the curtain, since asking Fable to show its work in the answer can quietly nudge you toward a weaker model. And if you are handing it Skills or templates built for last year’s models, loosen them first, because rigid old instructions drag this model down.

The honest warning: the perfect version takes far longer than a day. Pick the two workflows you are most tired of rebuilding, get them into Skills and/or Agents, and leave the rest for later. There is a setup tax here, an afternoon of describing how you work in precise language, and it feels slow while you are paying it. It pays itself back the first week you stop re-explaining the same context every Monday.

Everything you build today keeps working after the deadline. The Skills sit with you and travel between tools. Your daily drafting and reformatting still run fine on the cheaper models. Fable stays on the API for the hard jobs, and you reach for it on purpose, when the task earns the cost. The deadline marks the moment when extravagance starts costing money, making today the right day to be deliberate about what you ask it to build.

Go build one Skill before the clock runs out. If you get stuck on which one, my inbox is open.

— Lauren Eve Cantor

thanks for reading!

if someone sent this to you or you haven’t done so yet, please sign up so you never miss an issue.

I’ve also started publishing more frequently on LinkedIn, and you can follow me here

if you’d like to chat further about opportunities or interest in AI, or this newsletter, please feel free to reply.

banner images created with Midjourney.