You open Claude Code, ready to dive into a session. Then you see it: a message that stops you cold. “Credits needed.” Your heart sinks. Did you miss an email? Is your subscription still active? You check your plan—everything looks fine. But the message is there. And you’re not alone.
Over the past few days, a growing number of users have reported that Claude Fable—the model that was supposed to be included in your subscription—now demands usage credits. Some see it. Some don’t. The rollout is inconsistent. The company hasn’t announced anything. And that’s exactly the point.
The most expensive product change is the one you never announce.
This isn’t a bug. It’s not a rollout glitch. It’s a carefully calibrated experiment to see how much pain users will absorb before they push back. Anthropic is quietly A/B testing a shift from subscription-based access to usage-credit metering for Claude Fable. And they’re doing it without telling you, because they want to measure your tolerance, not your opinion.
One user on X noticed the change and posted a screenshot. Another replied: “I’m still using Fable 5 right now without usage credits.” A third reported seeing translated instructions: “Fable is further in your max plan. If you get messages ‘credits needed,’ just restart CC.” The confusion is the point. Ambiguity buys time. It lets the company observe real backlash, grandfather some users, and slowly normalize the new model as inevitable rather than optional.
Here’s the tension that makes this so dangerous: a subscription is a promise. You pay a fixed amount, you get predictable access. Usage-based pricing, on the other hand, is a relationship that never stops asking for money. When the product’s promise and its cost structure are fundamentally at odds, the product always loses. And the user always pays—either in money or in anxiety.
Subscription pricing is a trust contract. Usage-based pricing is a relationship that never stops asking for money.
If you rely on Claude Code or Fable for daily work, this matters. A lot. Your monthly costs could become unpredictable. Your workflow could be interrupted by a credit check. Your trust in the platform could erode, one quiet experiment at a time. And because the change is being tested live, there’s no official line to point to, no FAQ to consult. You’re left guessing whether you’re in the control group or the treatment group.
Anthropic’s economics favor usage-based pricing. That’s no secret. But the way they’re rolling this out—subtly, silently, with a split test—reveals something deeper. They know that if they announced the change outright, the backlash would be immediate and loud. So they’re letting the new normal creep in, one user at a time, until it’s too widespread to reverse. By the time they make it official, the question won’t be “Do you accept this?” It will be “Didn’t you already know?”
The quietest price change is the one that’s over before you realize it’s happening.
This isn’t a conspiracy theory. It’s a pattern we’ve seen before—from SaaS companies, from streaming services, from every platform that started with a simple promise and ended with a complex pricing page. The difference here is that Claude Fable is a tool you depend on, not a subscription you can cancel on a whim. The switching costs are high, and Anthropic knows it.
So what do you do? Start tracking your usage now. Keep screenshots of your current plan. If you see the “credits needed” message, document it. And most importantly, don’t assume this is a bug. It’s a test. And you’re the subject.
Don’t wait for the official announcement. The experiment is already running. The question is: are you in the control group or the treatment group? And more importantly—are you okay with being the lab rat?
FAQ
Q: Is this really an A/B test or just a bug?
A: It's almost certainly an A/B test. The inconsistent rollout—some users see it, others don't—is the hallmark of a controlled experiment. A bug would affect everyone or no one, not a carefully segmented subset.
Q: What should I do if I see the 'credits needed' message?
A: Document it immediately: screenshot, note your plan tier, and contact support. But don't expect a clear answer. The ambiguity is intentional. In the meantime, start tracking your usage to understand how much this change could cost you.
Q: Isn't usage-based pricing fairer for light users?
A: In theory, yes. But the problem is the bait-and-switch: you signed up for a subscription with a specific promise, and now that promise is being altered without consent. Fairness requires transparency. This experiment is the opposite of transparent.