Anthropic Separates Fable 5 From a Claude Code Usage Promotion
From July 20, Fable 5 remains included up to 50% of weekly limits on Max and premium seats; Pro and standard seats use credits. A separate temporary 50% increase, through August 19, applies only to Claude Code.
On July 19, 2026, the promotion that let every paid plan try Fable 5 at no additional cost ended. From July 20, Anthropic's official policy splits access by plan: Max and premium seats retain an included amount, while Pro and standard seats move to usage credits. The increase that lasts into August is a different promotion and applies only to Claude Code.
The distinction matters because “available on a plan” is not the same as “included at no extra charge.” On Max, Team premium, and legacy Enterprise premium seats, Fable 5 may use up to 50% of the weekly limit already purchased. On Pro, Team standard, and legacy Enterprise standard seats, using that model draws on pay-as-you-go credits from the first token. There was no uniform postponement for every subscriber.
The lasting lesson is not to guess whether flat pricing is ending, but to read five variables before comparing an offer: model, plan, product surface, promotional window, and limit type. Confusing just two of them here reverses what the published terms actually say.
What changes by plan
For Max, Team premium, and legacy Enterprise premium, Anthropic says Fable 5 remains included for up to 50% of weekly limits. That use does not create a second allowance: it consumes the same weekly plan limit faster. Once included usage is exhausted, a user can continue with credits if additional usage is enabled.
For Pro, Team standard, and legacy Enterprise standard, Fable 5 is not part of the included allocation from July 20. The plan-by-plan table says the model uses pay-as-you-go credits from the start of a session. Consumption-based Enterprise plans and API access are billed at standard rates.
Anthropic paired the change with a one-time credit promotion for eligible Pro and Team standard users: $100 per plan or seat, capped at $2,500 per Team organization. The credits had to be claimed between July 20 and August 2 and expire on September 17, 2026; they do not make Fable 5 a permanent plan benefit.
The official Fable 5 page lists API prices of $10 per million input tokens and $50 per million output tokens, with a 90% input-cache discount. Those units can price an API call; they cannot establish a monthly subscription bill without knowing what usage is included, which credits were enabled, and how many billable tokens were recorded.
The other 50%: a separate promotion
From May 13 through August 19, 2026, Anthropic provides a 50% increase in weekly Claude Code limits for Pro, Max, Team, and legacy seat-based Enterprise plans. The page confines it to Claude Code: Claude Chat and Cowork do not increase, five-hour session limits remain unchanged, and consumption-based Enterprise plans are excluded. After August 19, weekly Claude Code limits return to standard without changing plan billing.
Which limits still apply
The support pages reviewed describe eligibility, billing, and duration, but do not attribute the changes to a particular business cause. Compute capacity, profitability, or preparation for a financial transaction would be explanations outside those terms. Keeping the verifiable rule separate from a possible motive prevents a plausible inference from becoming an Anthropic statement.
Limits are not one fixed number for every conversation. Claude's usage guide says they vary with message and file length, complexity, features, model, and reasoning effort. Usage is shared across Claude on the web, desktop, and mobile, as well as Claude Code and Cowork, when they belong to the same account.
Two clocks are involved. The session limit resets every five hours; the weekly limit recovers seven days after its relevant window begins. Reaching one does not mean the other is exhausted. A separate context limit applies to each conversation. The Claude Code promotion expands only that surface's weekly allowance, so it cannot be described as 50% added to the entire plan.
Additional usage is optional and separate from the subscription fee. The credit documentation says it is prepaid, billed at standard API rates, and can carry a monthly cap, automatic reload, and alerts. Claude asks for confirmation before continuing beyond included limits. Turning the feature off prevents plan exhaustion from automatically becoming variable spending.
How to read price without mixing units
A subscription fee, a weekly limit, and a token price answer different questions. The pricing page lists Pro at $20 monthly and Max from $100, while warning that limits apply. The fee buys access and an allocation; per-token pricing determines the cost of consumption outside that allocation or through the API.
Even token pricing separates input, output, and caching. A request with substantial reused context can cost differently from another showing the same number of visible words. A calculation based on “words sent and received” therefore does not reproduce a bill; it needs the billable-token record and the rule applied to each category.
Available does not mean included
Fable 5 is described as available on Pro, Max, Team, and Enterprise, but the model page points users to plan-specific terms. It also warns that safeguards may route some prompts to another model and that Fable 5 conversations have 30-day data retention. Access, model served, retention, and billing are independent clauses.
“Up to 50% of weekly limits” does not promise that half is reserved for Fable 5 either. It means that model may occupy at most that portion of the shared weekly budget and consumes it faster. To learn what remains, users must check the usage indicator on their account, not add one assumed included 50% to another promotional 50%.
The matrix also prevents generalizing from a plan price. Two Team users can face different conditions if one holds a standard seat and the other a premium seat. The same applies to legacy Enterprise versus consumption-based Enterprise. The commercial product name alone does not identify the unit being billed.
What each user should check
A Max user should distinguish the shared weekly limit from the maximum portion Fable 5 can occupy. A Pro user should check whether promotional credits were claimed, when they expire, and whether additional usage is enabled. A team must inspect seat type. Anyone using Claude Code should keep its temporary increase separate from Chat and Cowork limits.
Dates also change the answer. The free Fable 5 trial ended July 19; the continuing terms began July 20; one-time credits could be claimed through August 2 and expire September 17; the Claude Code expansion ends August 19. A screenshot without a date may be accurate and, days later, describe a different policy.
Exposure to a variable bill does not follow simply from being an occasional or heavy user. It follows from the exact combination of plan, model, available credit, and controls. A heavy user with additional usage disabled can stop at the limit; one with automatic reload can keep spending. The useful setting is not a guess about the bill but a monthly cap and alerts before a long task begins.
The transferable skill is to translate an AI offer into a table before drawing conclusions: which model, on which plan and seat, inside which product, during what dates, under which limits, and what happens after they are crossed. Applied here, the table reveals two simultaneous policies, not a postponed general charge: Fable 5 has plan-specific rules, while Claude Code receives a separate temporary increase.
This article was produced with artificial intelligence under human editorial oversight.
