Claude Fable 5 Splits the Plans: Max Keeps It, Pro Gets $100 and a Bill

Last updated: July 20, 2026

I use Claude every day for productivity, catching grammar mistakes, building apps, etc. Because of that, I’ve been watching the Fable 5 situation closely. Anthropic pulled the model, brought it back, said it was leaving subscriptions, extended the deadline, extended it again, then extended it a third time. Five weeks of that.

Today it finally settled, and the answer splits paying customers into two groups. If you’re on Max(me), you keep Fable 5. If you’re on Pro, you don’t. Here’s exactly what changed, why I think it kept moving, and what the numbers mean for your bill.

What Changed With Claude Fable 5 Today

Starting today, Claude Fable 5 is a permanent subscription feature for Max and Team Premium subscribers, and it is no longer included for Pro and Team Standard. Anthropic announced the structure on July 17.

The details matter more than the headline, so here they are:

• Max and Team Premium: Fable 5 is included permanently, but capped at 50% of your standard weekly usage limits. Those standard limits also shrink on the same day, so the cap applies to a smaller number than it did last week.
• Pro and Team Standard: Fable 5 is out of the subscription. You get a one-time $100 usage credit, and once that runs down, Fable 5 bills at API rates: $10 per million input tokens and $50 per million output tokens.
• Everyone: all Claude models pull from the same weekly pool. Heavy Sonnet 5 or Opus 4.8 use eats into your Fable 5 headroom, and the reverse is true too.

Anthropic has not said when the $100 credit lands in accounts or what it expects a typical person to get out of it.

Why the Deadline Kept Moving

Anthropic moved the cutoff three times in five weeks, from July 7 to July 12 to July 19, and the original plan was more severe than where it landed. The company had intended to drop Fable 5 from subscription plans entirely and sell it only through per-token API billing.

The back-and-forth goes back further than the pricing fight. Fable 5 and Mythos 5 went offline on June 12 under a US export-control order, the Commerce Department lifted that order on June 30, and Anthropic restored Fable 5 on July 1 across Claude.ai, Claude Code, and the rest of its surfaces. So the model came back from a shutdown and immediately walked into a subscription question.

Each extension followed user pushback. Android Authority covered the extension that came after backlash over the early cutoff, and at one point Anthropic told BleepingComputer that Fable 5 was not permanently leaving subscriptions. That turned out to be true for Max and not true for Pro.

What GPT-5.6 Sol Has to Do With It

OpenAI’s GPT-5.6 Sol reached general availability on July 9 at $5 per million input tokens and $30 per million output tokens, which is half the input cost and 60% of the output cost of Fable 5. Reporting from TechTimes ties Anthropic’s reversal directly to that release.

The capability gap is thin. On the Artificial Analysis Intelligence Index, Fable 5 scores 59.9 and Sol scores 58.9, close enough to call a tie. The bigger number is what a finished task costs: Artificial Analysis found Sol completes equivalent work at roughly a third of Fable 5’s cost, because it uses fewer output tokens to get there.

Line those up and the retention math writes itself. A premium plan that doesn’t include your best model is a hard sell when the competition’s flagship costs less per task at nearly the same capability. I ran the Sol versus Fable 5 numbers earlier this month, before any of this was settled, and the pricing gap was the story then too.

What the $100 Credit Actually Buys

Less than it sounds like. At $50 per million output tokens, a single session that generates 2 million output tokens costs exactly $100, and that is not an extreme session for the kind of work Fable 5 is built for.

An autonomous multi-file coding migration or a deep document analysis run can hit that in one sitting. Fable 5 is designed for long, iterative generation, which is precisely the workload that burns output tokens fastest. So the credit is sized against the per-token rate rather than against how people actually use the model.

For a Pro subscriber doing light work, $100 might last a while. For a Pro subscriber who leaned on Fable 5 for heavy jobs, it’s closer to a single good session than a monthly allowance.

What to Do, Depending on Your Plan

The right move depends on how much you actually use Fable 5, not on which plan sounds better. Check your usage before you change anything.

• On Max or Team Premium: nothing to do today. Watch the 50% cap, because your total weekly limits also dropped, and the cap is half of the new smaller number.
• On Pro, light Fable 5 user: take the $100 credit and carry on. Opus 4.8 and Sonnet 5 are still included and handle most everyday work without touching the credit.
• On Pro, heavy Fable 5 user: you have three options, and they’re all real. Upgrade to Max, drop down to Opus 4.8 for the same tasks, or budget for API billing after the credit runs out.
• Shopping around: GPT-5.6 Sol is the obvious comparison right now on price per task. My breakdown of which AI is worth paying for covers how I split work across tools instead of betting everything on one.

My honest read on the whole thing is this looks like a company doing everything it can to stay competitive, not a company that made up its mind and stuck to it. GPT-5.6 Sol is a genuinely capable model at a lower price, and Anthropic moved the deadline three times before landing somewhere that keeps its best model in front of its highest-paying customers. The race at the top is tight right now, and Pro subscribers ended up as the part of the ledger that got adjusted.

Five weeks, three extensions, one split. Check which side of it you’re on before your next big session.

Related reading: GPT-5.6 Sol vs Claude Fable 5: the benchmarks are tied, the prices aren’t | Which AI should own which part of your workflow | New to AI? Start here