AI Model Report

Reviews · JULY 19, 2026

Anthropic Re-Tiers Fable 5: Max Keeps It, Pro Gets a $100 Wallet

Starting July 20, Claude Fable 5 is bundled into Max and Team Premium at 50% of weekly limits — which themselves shrink by a third the same day — while Pro and Team Standard drop to a one-time $100 credit at API rates.

By Karl Strauchman · Senior model reviewer · July 19, 2026

On July 17, Anthropic announced that Claude Fable 5 will settle into a permanent, and permanently bifurcated, place on its price sheet. Starting July 20, Fable 5 is bundled into Max and Team Premium at 50% of weekly usage limits. Pro and Team Standard subscribers lose seat-based access entirely and receive a one-time $100 credit metered at API rates: $10 per million input tokens and $50 per million output.

The wrinkle sits underneath the headline. Those weekly caps themselves fall by roughly a third on the same day, as the temporary Claude Code rate-limit boost expires. Half of a smaller pie is what Max buyers are actually getting.

The redeployment ends six weeks of subscription limbo. Fable 5 launched June 9 alongside Mythos 5, its safety-classifier-free sibling. Three days later, a US export-control directive pulled both offline after Amazon researchers reportedly bypassed the safeguards. Anthropic redeployed Fable 5 on July 1 at 50% of limits across Pro, Max, Team, and select Enterprise plans; that window closed July 7, with access extended on credits while the classifier retrained. Enterprise Premium lost Fable 5 access the same day. The July 17 "Redeploying Claude Fable 5" post is Anthropic's attempt to convert a rolling operational patch into a stable product SKU.

It's also a quiet concession to the economics. Fable 5 is expensive to serve, and Anthropic's own pricing ladder makes that legible: Haiku 4.5 at $1 / $5, Sonnet 5 at $2 / $10 through August 31, 2026, Opus 4.8 at $5 / $25, and Fable 5 at the top at $10 / $50. Handing Pro users a $100 wallet at those rates isn't a downgrade dressed up as a perk; it's the honest version of what unmetered access was always going to become.

The competitive picture sharpens the move. On the Artificial Analysis Intelligence Index as of July 18, Fable 5 sits at roughly 59.9 to GPT-5.6 Sol's 58.9, effectively a tie. But Sol is priced at $5 / $30, and because it emits fewer output tokens, its equivalent-task cost lands at roughly a third of Fable 5's. OpenAI is selling the same intelligence tier for meaningfully less, and Anthropic knows it.

Hence the split. Max becomes the tier where you pay a subscription premium for the frontier model as an entitlement. Pro becomes a metered relationship with the API, sold through a wallet. The middle disappears. That's a familiar shape for anyone who watched cloud infrastructure pricing settle in the mid-2010s, when flat-rate compute quietly gave way to consumption billing wherever the workloads got expensive enough to matter. Frontier inference has now crossed that threshold.

Sources