Anthropic Finally Ends the Fable 5 Deadline Loop — Splitting Claude Into a Permanent Two-Tier Model, and Pro Users Aren't Happy

The Three-Extension Saga Ends With a Permanent Split, Not a Fourth Deadline

On July 20, 2026, Anthropic closed the loop this blog tracked through three extensions — July 7, then July 12, then July 19 — by skipping a fourth deadline entirely. In an announcement titled "Redeploying Claude Fable 5," the company confirmed that instead of another temporary reprieve, Fable 5 access now permanently splits by plan tier: Max and Team Premium subscribers keep the model built into their subscription at 50% of their weekly usage limit, indefinitely, while Pro and Team Standard subscribers lose free inclusion outright. Developer and AI writer Simon Willison flagged the shift two days early, on July 18, after Anthropic previewed the permanent-tier plan ahead of the formal cutover.

What Each Tier Actually Gets, In Dollars

For Max and Team Premium seats, nothing about the terms changed from the prior extensions — Fable 5 stays included at half the weekly usage pool, but the word "permanent" replaces "until [date]" in Anthropic's documentation. Pro and Team Standard subscribers get a one-time $100 usage credit to spend on Fable 5, and once that's exhausted, every additional token bills at API rates: $10 per million input tokens and $50 per million output tokens, the steepest per-token price in Anthropic's entire lineup. For a Pro subscriber running Fable 5 through Claude Code on anything beyond light use, $100 in credits can disappear inside a single day of agentic coding sessions.

Why a Permanent Split Beats a Fourth Extension

The prior three deadlines each followed the same pattern: an announced cutoff, visible backlash, then a walk-back days or hours before the deadline hit. A fourth extension would have kept that cycle running indefinitely and told subscribers nothing about what to actually plan around. A permanent two-tier structure resolves the ambiguity in Anthropic's favor without requiring another public reversal — it also sidesteps the question this blog raised on July 19, since a structural pricing decision doesn't need to wait on whether an Opus 5-class model built on the leaked "Honeycomb EAP" specs ships before or after any given date.

"We Got 7 Days of the 14 We Were Promised" — The Pro-Tier Backlash

PCWorld's headline on the reaction was blunt: "Claude subscribers are furious over Fable's new restrictions." A widely shared Reddit thread captured the core complaint from Pro users who'd budgeted around the original promotion: "Not a good look to bring Fable back and then both halve the usage and take away days," one user wrote, with another adding, "We got to use it for like 3 days out of the 14 we were told, and now we get it for just 7 days at half usage?" On Hacker News, the tone skewed toward reconsidering the relationship entirely, with one commenter writing, "Considering the cash my organization dumps on them, we hoped for better... We're all pretty leery now and investigating other models."

The GPT-5.6 Sol Math Hasn't Changed

The competitive backdrop this blog covered on July 19 is still the crux of the pricing decision. GPT-5.6 Sol prices at $5 per million input tokens and $30 per million output tokens — roughly half of Fable 5's now-permanent API rate — while scoring within a point of Fable 5 on Artificial Analysis's Intelligence Index. Every Pro subscriber who burns through their $100 credit and switches to API billing is now making that same cost-versus-capability comparison Anthropic had been avoiding by extending free access; the permanent split doesn't change the math, it just forces the decision onto a fixed schedule instead of a moving one.

What to Check Before Your Team's Next Claude Code Sprint

Two concrete steps for engineering teams. First, confirm your actual plan tier today — Max and Team Premium keep Fable 5 indefinitely at no extra cost, while Pro and Team Standard are now on a $100 clock that agentic, multi-step Claude Code sessions can burn through fast. Second, if your team sits on Pro or Team Standard and relies on Fable 5 for coding work, model the $10/$50 API rate against GPT-5.6 Sol's $5/$30 now, before the credit runs out and the switch happens under deadline pressure instead of by choice.

Bottom Line

Three deadlines and five weeks of "access whiplash" ended not with a resolution developers wanted, but with a permanent structure that formalizes the split this blog has tracked since July 14: Max keeps Fable 5 free, Pro pays per token after $100 runs out, and the backlash on Reddit and Hacker News suggests plenty of Pro subscribers feel it as a downgrade rather than a settlement. With Opus 5 still unconfirmed and GPT-5.6 Sol sitting at half the price for comparable capability, the permanent split may turn out to be less of an ending than the moment Pro-tier developers start shopping elsewhere.