Anthropic Launches Claude Fable 5.2: Cheaper API, Just One Week for Pro Users

Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on 9 November 2026, pitching them at heavy technical work in coding, finance and science. For most people the real story is access, not capability: the Free plan gets nothing, Pro subscribers get Fable inside their plan for one week only (until 16 November), and Mythos is closed to ordinary users altogether. Our advice: if you pay for Pro, use this week to test Fable on your own work before you switch on extra usage credits.

What changed

According to Anthropic’s announcement, Fable and Mythos are the same underlying model. Fable carries strict cybersecurity and biology safeguards; Mythos ships without them. Anthropic says both return answers up to 30% faster than the 5.1 versions and improve on coding, reasoning and vision. The Verge reports that Anthropic positions Fable 5.2 for hard software engineering and for orchestrating teams of agents, which matters most if you work in Claude Code.

Developers get a price cut. TechCrunch notes that Fable’s input price falls from $10 to $8 per million tokens; output is $40 per million for both models. The API model ID is claude-fable-5-2, available on the Claude Platform, AWS, Google Cloud and Microsoft Azure. Fast mode costs twice as much: $16 input and $80 output per million tokens.

Who gets Fable 5.2, and what it costs

Here is what Anthropic’s pricing page and its launch-promotion help article say:

• Free: no Fable.
• Pro ($20 a month, or $17 a month billed yearly) and standard Team seats: Fable normally runs on separate, paid usage credits. During the launch week, until 16 November 2026 at 11:59 pm PT, it counts against your normal weekly limit instead, capped at 50% of that limit. Accounts that were paying before 9 November at 10:00 am PT also get one “Fable reset”, which expires when the week ends.
• Max ($100 a month) and Ultra ($200 a month): Fable is included up to half of the weekly limit. Subscribers get one to three resets depending on tenure (under six months: one; six to twelve months: two; over a year: three), valid until 8 March 2027.

A reset does not raise your limits. It gives back the Fable usage you have already spent that week. Unused promotional allowance does not carry over.

Mythos 5.2 is not for regular users

No consumer plan lets you pick Mythos. Anthropic limits it to vetted organisations that apply through its Cyber Verification Program or Life Sciences Verification Program, and Claude Enterprise customers can also use it for code scanning through Claude Security. Fable, by contrast, hands blocked questions to an older model: Opus 4.8 answers when a cybersecurity query is refused, Opus 5 when a biology one is. Anthropic says it has cut false refusals since Fable 5.1.

The launch also lands amid safety questions. TechCrunch points to a METR report that found Anthropic’s internal models were speeding up the company’s AI research by roughly 2x, which raised recursive self-improvement concerns. Anthropic says it deliberately limited the research skills that could most directly accelerate AI development.

What about Arabic?

Anthropic’s announcement says nothing about Arabic and publishes no Arabic results, so don’t assume better coding means better Arabic writing. On availability, Anthropic’s supported regions list includes Saudi Arabia, the UAE, Egypt, Jordan, Morocco, Qatar, Kuwait, Oman, Bahrain, Tunisia, Algeria, Iraq, Lebanon, Libya, Palestine, Sudan and Mauritania. Syria and Yemen are not on it. Prices are shown only in US dollars, with no regional pricing.

What to do now

If you are on Pro in a supported country, run a short Arabic test before 16 November. Give Fable 5.2 and the model you normally use (Opus 5.5, for example) the same input:

• Dialect to standard Arabic: paste a paragraph in Gulf or Egyptian dialect and ask for a clean Modern Standard Arabic version. Check the grammar and whether the meaning survived.
• Arabic code comments: ask for a short function with Arabic comments and see whether right-to-left text tangles with the code or breaks the punctuation.

If you can’t see a clear difference, there is no reason to turn on paid credits after the week ends. For a broader view, see our Claude AI page and our ChatGPT vs Claude and Claude vs Gemini comparisons.