Anthropic has announced another extension of Claude Fable 5 availability in Claude Max plans: now through July 19. The announcement comes days after OpenAI’s release of GPT-5.6 Sol, and confirms that Fable remains in the same capability tier (Mythos).

Why it matters. As we reported on July 5, Anthropic had justified Fable access restrictions by citing compute constraints: they wanted to gauge demand and availability before committing. OpenAI, by contrast, placed no such limits on GPT-5.6. This third extension (following previous ones) signals that Anthropic still sees value in keeping Fable accessible to Max customers — likely because the model holds its own against Sol on everyday tasks, even as competitive pressure remains high.

Among those tracking the iterations closely, some suggest Anthropic should change course and keep Fable permanently available in Max plans: uncertainty around dates is driving users to OpenAI.

The concrete signal. If you use Claude Max for Fable-based work, you have roughly another week at current conditions (half the available weekly limit for Fable). After July 19, you can continue using Fable with usage credits — but the model won’t disappear entirely as previously feared.

In detail

Context: from compute constraints to competitive pressure

When Anthropic released Claude Fable 5 in May, it immediately clarified that Max plan access would be temporary: too much uncertainty around demand and compute availability to commit long-term. The model came with a cap (half the weekly limit), and with an expiration date Anthropic has already shifted three times.

On July 5, after weeks of suspension for US export compliance checks, Fable came back online. On July 9, OpenAI launched GPT-5.6 with three variants (Luna, Terra, Sol), all in the same Mythos/Fable tier. Sol in particular arrived with no access restrictions for Plus, Pro, and Business plans — a signal that OpenAI feels confident handling the demand.

Anthropic today shifts the date again, this time to July 19, and adds that Claude Code limits (the integrated development environment) also remain 50% higher than baseline.

What sources say

Anthropic’s official X announcement confirms the extension and maintains the formula: «up to half the weekly limit for Fable 5, then usage credits or switch to another model». OpenAI, meanwhile, is gaining ground precisely on uncertainty: those planning work prefer a guaranteed model to one that might vanish in a week.

Thibault Sottiaux (OpenAI) stated yesterday they temporarily removed the 5-hour limit for Plus, Business, and Pro plans, and are optimizing GPT-5.6 Sol to consume less. The implicit message: OpenAI has no compute issues with Sol.

The implications

If Anthropic keeps shifting the date instead of committing, the signal to customers is ambiguous: Fable is good enough to keep, but not good enough to guarantee. For those using Claude Max in production, this means planning with a backup plan. For those evaluating Claude versus ChatGPT, access stability matters.

Fable’s technical performance isn’t in question: recent benchmarks confirm it’s competitive with Sol. The issue is trust: a model that tomorrow could cost triple or require extra credits is harder to embed in a stable workflow.

What’s missing

Anthropic hasn’t yet said what happens after July 19. The most likely scenario is another extension (this is the third), but without a clear announcement on permanent availability or final pricing, the message to customers remains «use Fable while it lasts».