Fable 5.2 / Mythos 5.2

Claude Fable 5.2 and Claude Mythos 5.2 are Anthropic’s 9 November 2026 upgrade for the models aimed at the hardest technical work, with claimed gains in coding, reasoning, and vision. They are the same underlying model. Fable ships with stronger cybersecurity and biology safeguards, which is why it is available to all users. Organizations doing beneficial work in cybersecurity or biomedical research can apply for Mythos through Anthropic’s expanded Life Sciences and Cyber Verification Programs.

List pricing is $8 per million input tokens and $40 per million output tokens, described as lower than the previous generation (the radar excerpt recorded that as about a 20% cut). Anthropic says the models generate outputs up to 30% faster than Claude Fable 5.1 and Claude Mythos 5.1. An optional fast mode speeds token generation by up to 2.5 times at $16 per million input tokens and $80 per million output tokens. Developers can start with claude-fable-5-2 on the Claude Platform. Fable 5.2 is available on all platforms, including Amazon Web Services, Google Cloud, and Microsoft Azure.

Highlights:

• Official post: https://www.anthropic.com/claude-fable-and-mythos-5-2 ; fuller safety and eval detail is in the System Card;
• For most everyday coding and knowledge work, Anthropic still recommends starting with Claude Opus 5.5, then moving up to Fable or Mythos—or down to Sonnet 5.5 or Haiku 5.5—based on scope;
• Fable 5.2 can also act as an Advisor (consulted by Haiku, Sonnet, or Opus on harder problems) or an Orchestrator (delegating simpler work to smaller models);
• Eligible paid plans received extra Fable usage during launch week; Mythos and Mythos fast mode stay limited to verification-program organizations.

Why it matters

This is a same-day upgrade of Anthropic’s tier for the most ambitious, open-ended technical work. Official examples include a team of about 1,000 Fable 5.2 agents formalizing the proof of Fermat’s Last Theorem in 30 hours (Anthropic says that used about one-third as many action tokens as an earlier research run and produced a proof about seven times shorter); coordinating more than 1,000 subagents over eight days to write a prototype operating system that boots on a laptop; designing three working electronic devices from a one-paragraph prompt with hardware partner Diode (a home voice-assistant hub, a USB speakerphone, and a handheld virtual pet); and completing two open calculations in black-hole / holographic-duality physics. Mythos 5.2 is described as Anthropic’s most capable model for cybersecurity engineering and is also used on life-science tasks such as viral capsid structure prediction. Claude Enterprise customers can use Claude Security, now including Mythos 5.2, to scan codebases.

Anthropic’s benchmark table says Fable 5.2 and Mythos 5.2 sit a step above Opus 5.5 on almost every capability they evaluated—for example 49.1% on FrontierCode 2.0 at Xhigh effort, 74.2% on Humanity’s Last Exam with tools, and 89.3% on OSWorld 2.1 (partial). Those figures come from the company post and System Card. Effort levels, safeguard settings, and internal evals matter; treat them as Anthropic’s reported numbers, not independently reproduced scores.

Limits and safety

Fable and Mythos are not meant to replace a daily-driver model. Fable’s classifiers block high-risk cyber activity (including penetration testing, exploit generation, and binary-based vulnerability work) and most research-biology tasks; blocked cyber queries fall back to Opus 4.8 and biology queries to Opus 5. Anthropic says Fable 5.2 produced about 55% fewer unnecessary blocks than 5.1 (about 60% fewer in cybersecurity and 40% fewer in biology), but many security-research workflows will still need Mythos. Pre-release safety work included alignment training, new sandbox-escape evaluations, and external testing by the UK AI Security Institute, Frontier Design, and METR; Anthropic assesses high-stakes misalignment risk as remaining low. Enterprise Frontier Safeguards—which can run automated safety monitoring on prompts and responses stored in a customer’s own infrastructure, compatible with zero data retention—is in private beta with customers such as Bridgewater, Salesforce, and Stripe. Pricing, verification applications, and launch-week usage promotions follow the official pages at the time you check them.

1. Read https://www.anthropic.com/claude-fable-and-mythos-5-2 and the System Card, and decide whether you need generally available Fable 5.2 or verification-gated Mythos 5.2;
2. Select claude-fable-5-2 in the Claude apps, Claude Code, or the Claude Platform; it is also on AWS, Google Cloud, and Microsoft Azure. If you want higher throughput, check where fast mode is offered (Anthropic lists Claude Code, the Claude Platform, Claude Models on Google Cloud, and Anthropic-hosted Microsoft Foundry) and the $16/$80 price;
3. Keep Opus 5.5 as the default for most daily work and reserve Fable for harder, longer technical jobs; follow Anthropic’s docs if you want it as an Advisor or Orchestrator;
4. Cybersecurity and life-sciences organizations should apply through the Cyber or Life Sciences Verification Programs for Mythos. Read the safeguard notes before assuming Fable will run a dual-use workflow the classifiers are designed to block.