Available Is Not the Same Word as Included

Anthropic shipped Claude Fable 5.1 on September 1. It is better than what came before it. Getting your hands on it is a separate question, and so is what it costs you once you do.

Anthropic lists Fable 5.1 as available on every paid plan, which leaves the free plan out. On Max it comes out of the allowance you already bought, capped at half your weekly limit. On Pro, the cheapest paid plan, it bills separately, on top of the subscription, at API rates.

The Ladder, as Anthropic Published It on September 6, US List Prices

These terms have already moved once this year. Everything below is what Anthropic published on September 6.

Anthropic's pricing page shows only "From $100" for Max. The $200 for Max 20x comes from its Max plan help article, updated August 7, 2026.

On Max, Fable is included up to half of your weekly usage limit. That half comes out of the allowance you already pay for. Anthropic's own words are that Fable models "draw from your plan's regular weekly usage limits and use them faster than other Claude models," so you hit the wall sooner. Anthropic sells Max 5x as five times more usage per session than Pro. Sessions are how it publishes that number.

A standard Team seat is $20 per seat per month billed annually, and it handles Fable the way Pro does, on credits. A premium seat is $100 per seat per month billed annually and includes Fable under the same half-your-weekly-limit ceiling as Max. Team is a two-seat minimum, so the cheapest Team setup that includes Fable is $200 a month billed annually.

What the Cheapest Paid Tier Actually Buys

On Pro, Fable 5 and Fable 5.1 sit outside the plan's usage limits, so your $20 buys the right to buy Fable and not one token of Fable usage. Those credits bill at standard API rates, which for Fable 5.1 is $10 per million input tokens and $50 per million output.

Opus 5, Sonnet 5 and Haiku 4.5 all sit inside the allowance, along with Claude Code and Claude Cowork. The free plan gets Sonnet and Haiku, with no Opus.

Opus 5 is in the $20 plan, and Anthropic's own developer docs say to start most work there, moving up to Fable 5.1 only when your own tests fall short on the hardest reasoning jobs or on work that runs for hours without you. That is written for developers picking a model in the API. What an owner can take from it is that Anthropic starts most work on the cheaper model too.

What It Looked Like On Our Own Work

We put Fable 5.1 on coding a couple of days after it shipped. Two days in, on September 5, it felt faster to us and less forgetful. Anthropic's own spec sheet rates Fable 5.1's latency "Slower" and Opus 5's "Moderate." It did not feel slower to us. Two days on one kind of work against a published spec sheet is not much to go on.

It has stayed on code since. The tokens run out pretty quickly and we would rather get it all in where it pays, so nothing else gets it.

What We Would Tell an Owner

Anthropic sells Fable 5.1 for big coding projects and for work that runs for days without you touching it. If none of your week looks like that, the plans that include it start at $100 a month and buy you a model you will never come close to using. For most of the owners we work with, the $20 plan is the right answer.

Claude Mythos 5.1, the other name in the September 1 announcement, is the same model with different levels of safeguards, and it is invitation only, so it is not something an owner can go out and buy.

So spend a month on Pro, do real work on Opus 5, and watch for the week you actually run out. If that happens on a job that mattered, you will know exactly what you are buying and why, which is a better place to make a $100 decision from than a launch announcement. If it never happens, the eighty dollars a month stayed in your business and you never had to spend an afternoon deciding.