Anthropic Launches Claude Fable and Mythos 5.2 AI Models

Anthropic released Claude Fable 5.2 to the public on Nov. 9, 2026, and opened Claude Mythos 5.2 to a limited set of vetted organizations, according to The Verge. The company says both models beat their 5.1 predecessors, released in September, on coding, reasoning, vision, and output speed.

What is Claude Fable 5.2, and who can use it?

Fable 5.2 is live for all users starting today, Anthropic told The Verge. The company positions the model for "daunting" technical work — software engineering problems, orchestrating teams of AI agents, and running long, multi-step projects without constant human steering.

What is Claude Mythos 5.2, and why isn't it open to everyone?

Mythos 5.2 is restricted at launch to organizations Anthropic has vetted, the company said, without detailing the screening criteria or a timeline for broader release. That staged rollout mirrors how frontier labs have handled earlier high-capability releases, giving enterprise partners early access while the company gathers real-world performance data.

What does Anthropic claim the new models can do?

During prerelease testing, Anthropic said the models "helped our engineers complete otherwise-infeasible software migrations, designed three working electronic devices from scratch, solved two open problems in the physics of black holes, made substantial progress on robotics evaluations, and formalized the proof of Fermat's Last Theorem in 30 hours," according to the company's statement reported by The Verge.

"Helped our engineers complete otherwise-infeasible software migrations, designed three working electronic devices from scratch, solved two open problems in the physics of black holes." — Anthropic, via The Verge

By the numbers

• 30 hours: time Anthropic says Fable 5.2 took to formalize a proof of Fermat's Last Theorem during testing.
• 2: open theoretical physics problems in black hole research Anthropic says the model helped solve.
• 3: working electronic devices the company says were designed from scratch using the model.

The Verge's report does not include independent verification of these claims, benchmark scores, or pricing details for either model.

How does this fit into the wider AI subscription market?

Anthropic's staged access model — broad release for Fable, vetted-only for Mythos — lands as rival labs experiment with their own tiered offerings. Meta, for instance, has rolled out AI-focused subscription plans aimed at consumer and creator users, a separate product strategy covered by HTT News. Hardware makers are also chasing the same demand curve; at a Saudi exhibition, DAREU unveiled adaptive gaming peripherals built around AI-assisted customization, per HTT News.

What should developers and enterprise buyers watch next?

• Whether Anthropic publishes benchmark data or third-party audits backing the physics and robotics claims.
• Timing and criteria for broader Claude Mythos 5.2 access beyond the initial vetted group.
• Pricing and API availability for Fable 5.2 inside existing Claude plans.
• Whether competing labs respond with matching claims on agent orchestration or long-running task completion.

Anthropic did not immediately specify pricing tiers or a public release date for Mythos 5.2 beyond the vetted-access phase, per the original report.

mindfulAI Keyboard — An iPhone keyboard that rewrites a draft in a tone you pick. Four free AI actions a day.