Anthropic launches Claude Fable 5.2 with lower API pricing

What did Anthropic launch on Nov. 9, 2026?

Anthropic announced Claude Fable 5.2 and Claude Mythos 5.2 on Nov. 9, 2026, describing them as new models for demanding technical work such as coding, reasoning and vision. The company said the two releases share the same underlying model, but are packaged differently for general availability versus restricted access.

Anthropic said Fable 5.2 is the broadly available version, while Mythos 5.2 keeps lighter cybersecurity and biology safeguards and is limited to approved security and life-sciences researchers and qualifying enterprise customers. The company also said organizations seeking beneficial cybersecurity or biomedical applications can apply for Mythos access through its verification programs.

How are Fable 5.2 and Mythos 5.2 different?

Anthropic says Fable 5.2 and Mythos 5.2 use the same model and core specifications, but the access rules differ. Fable 5.2 is generally available through the Claude platform and listed cloud marketplaces including Amazon Web Services, Google Cloud and Microsoft Foundry, with access for Pro, Max, Ultra, Team and Enterprise users.

Mythos 5.2 is active but requires verification through Anthropic programs such as its Cyber Verification Program. CNBC reported that Fable 5.2 was available at launch, while Mythos 5.2 was offered to vetted organizations through Anthropic’s Life Sciences and Cyber Verification Programs. The Verge likewise reported that Mythos 5.2 initially remained limited to vetted organizations.

Anthropic said Claude Enterprise customers can access Claude Security, which now includes Mythos 5.2.

What are the pricing and usage terms?

Anthropic cut the listed API price for Fable 5.2 to $8 per million input tokens and $40 per million output tokens, down from $10 and $50 for Fable 5.1. The company also said an optional fast mode is available at $16 per million input tokens and $80 per million output tokens.

Its platform documentation says Fable 5.2 has a 1 million-token context window, a maximum output of 128,000 tokens, a June 2026 knowledge cutoff and a default high-effort setting with adaptive thinking always on. Anthropic also said Fable 5.2 requires 30-day data retention by default and is not available under zero data retention unless expressly authorized, though TechCrunch reported the model is compatible with zero data retention policies through Enterprise Frontier Safeguards.

Anthropic said Batch API requests for Fable 5.2 are discounted 50% on both input and output, and prompt caching is priced at $10 per million tokens for 5-minute cache writes, $16 per million tokens for 1-hour cache writes and $0.20 per million tokens for cache reads.

What performance claims did Anthropic make?

Anthropic said Fable 5.2 can generate outputs up to 30% faster than Fable 5.1, although that claim was attributed to Anthropic. The company also said the models performed well on coding, scientific research, cybersecurity, protein-design and agentic-search evaluations.

In its announcement, Anthropic highlighted several internal examples meant to show the models’ technical reach. The company said an earlier prerelease version of Fable 5.2 migrated a critical production service from Python to Rust and made startup about 26 times faster. It also said one engineer had the model design a prototype operating system, with the model coordinating more than 1,000 subagents for eight days and producing a working system that boots on a laptop.

Anthropic said a 1,000-agent Fable 5.2 team formalized Fermat’s Last Theorem in 30 hours with minimal human intervention, while the company’s first formalization using an unreleased research version of Claude took 11 days. It also said Fable 5.2 solved two open problems in theoretical physics concerning black holes and quantum gravity, and that each of three hardware boards was ready for manufacturing in under a day with minimal human input.

For robotics, Anthropic said Fable 5.2 scored 64.9% on RoboDojo Sim at Max effort, compared with 39.4% for Opus 5.5 and 27.6% for Fable 5.1. The company also positioned Opus 5.5 as the recommended starting point for most coding and knowledge tasks, with Fable or Mythos for more ambitious work.

Why does Mythos 5.2 matter for security and life sciences?

Anthropic designed Mythos 5.2 for approved researchers and enterprise users working on cybersecurity and biology applications. The company said Mythos 5.2 found 10.2% more vulnerabilities than Opus 5.5 and 14.4% more than Mythos 5.1 on its internal CVE Rediscovery evaluation, which covered 1,829 CVEs in an October 2026 run.

On the life sciences side, Anthropic said Mythos 5.2 developed a method to predict whole AAV capsid structures and accurately modeled 28 of 34 capsids it attempted in the company’s tests. The model documentation says Mythos 5.2 is not scheduled for retirement before Nov. 9, 2027.

The release comes as Anthropic continues to expand its enterprise and cloud distribution. Anthropic said Fable 5.2 is available through the Claude platform and major cloud marketplaces, while the documentation says Fable 5.2 is scheduled for retirement no earlier than Nov. 9, 2027. CNBC reported that Anthropic generated $33.24 billion in revenue for the nine months ended Sept. 30, up roughly 1,200% year over year, and that the company is aiming for a $1 trillion to $1.5 trillion IPO valuation, down from a previously floated $2 trillion target.