Claude Fable 5.2 & Mythos 5.2: What's New, Price, Access

Anthropic just shipped its strongest models yet, and the internet is losing it. Claude Fable 5.2 and Claude Mythos 5.2 went live on 9 November 2026, and within hours the launch was sitting at the top of Hacker News. The flex list is wild: a proof of Fermat's Last Theorem formalized in 30 hours, two open black-hole physics problems solved, and three working gadgets designed from a one-paragraph prompt.

But what is actually new, what does it cost, and can you use it? Here's the no-hype breakdown.

TL;DR

Claude Fable 5.2 and Mythos 5.2 are the same underlying model. Fable has strict cyber and bio safeguards; Mythos doesn't, so it's limited to vetted organisations.

Fable 5.2 is available to everyone today on the Claude Platform, AWS, Google Cloud and Microsoft Azure/Foundry.

Price drops to $8 per million input tokens and $40 per million output tokens. Optional fast mode (up to 2.5x faster) costs $16 / $80.

Anthropic says output is up to 30% faster than the 5.1 models, with a 1M-token context window.

Anthropic still says most people should start with Claude Opus 5.5 and move up to Fable only for the hardest work.

What is Claude Fable 5.2?

According to Anthropic's launch post, Fable and Mythos are models "designed to tackle the most daunting technical challenges". Version 5.2 brings improvements in coding, reasoning and vision, and the two models follow Fable 5.1 and Mythos 5.1, which arrived in September (The Verge).

Think of Fable as Anthropic's "big brain" tier. It sits above Opus 5.5 (the everyday driver) and is meant for long, open-ended projects: huge code migrations, research and orchestrating teams of AI agents. New to agents? Our AI agents explainer covers the basics.

Anthropic's official Claude Fable 5.2 and Mythos 5.2 announcement image

Fable 5.2 vs Mythos 5.2: what's the difference?

Same model, different guardrails.

Claude Fable 5.2: available to all users. Classifiers block high-risk cyber work (penetration testing, exploit generation) and most research-biology tasks. When a request is blocked, Anthropic routes it to an older Opus model instead.
Claude Mythos 5.2: the same model with more permissive safeguards in those areas. Access is only through Anthropic's Cyber Verification Program or Life Sciences Verification Program. Claude Enterprise customers also get it inside Claude Security, Anthropic's code vulnerability scanner.

Anthropic says Fable 5.2's safeguards are as strong as 5.1's but trigger far fewer false alarms: about 55% fewer blocks in early testing (around 60% fewer in cybersecurity and 40% fewer in biology).

Claude Fable 5.2 benchmarks: the numbers that matter

Here are some of the scores Anthropic published. These are the company's own results, so treat them as claims until independent testers check them.

Benchmark	Fable 5.2	Fable 5.1	Opus 5.5	GPT-6 Astra
FrontierSWE v2 (agentic coding)	69.9%	56.3%	62.3%	65.5%
FrontierCode 2.0 (agentic coding)	49.1%	32.4%	45.5%	42.8%
Humanity's Last Exam (with tools)	74.2%	65.6%	67.7%	57.2%
OSWorld 2.1 (computer use)	89.3%	83.5%	87.2%	79.5%
RoboDojo Sim (robotics)	64.9%	27.6%	39.4%	29.8%

Not a clean sweep, though. On AutomationBench (business workflows), Google's Gemini 4 Carbon scored 53.9% against Fable 5.2's 51.4%.

Anthropic also says that at High effort, Fable 5.2 roughly matches Opus 5.5's best FrontierCode score for about half the cost per task. Want to see how the big chatbots compare for everyday use? Check our ChatGPT vs Claude vs Gemini guide.

The flex list: what Fable 5.2 actually did in testing

Anthropic's demos are the reason this launch went viral:

Fermat's Last Theorem in 30 hours. A team of about 1,000 Fable 5.2 agents fully formalized the proof in the Lean language in 30 hours. Anthropic's first attempt with an unreleased research model took 11 days.
Two open physics problems. Fable 5.2 solved two open problems about black holes in theoretical physics. Anthropic has published preprints on both.
Three working gadgets. From a one-paragraph prompt, it designed a home voice-assistant hub, a USB speakerphone and a handheld virtual pet, working with circuit board maker Diode. All three worked without electrical fixes.
A homemade operating system. One engineer had Fable 5.2 build a prototype OS from an empty folder. Over eight days it coordinated more than 1,000 subagents and produced an OS that boots on a laptop.
Python to Rust. Agents running an earlier version of the model migrated a production service from Python to Rust. Anthropic says it now starts about 26 times faster.

Anthropic also says more than 90% of the changes merged into its core codebase are now written by Claude (with human guidance). That stat alone is going to start a few arguments.

Mythos 5.2 and cybersecurity

Anthropic calls Mythos 5.2 its most capable model for cybersecurity engineering. On its internal CVE Rediscovery test (1,829 known vulnerabilities), Mythos 5.2 found 10.2% more than Opus 5.5 and 14.4% more than Mythos 5.1. A tester who has spent the year hardening Firefox with Claude told Anthropic that Mythos 5.2 "roughly doubled" Mythos 5.1's vulnerability discovery metrics in their scanning campaigns.

That's why it's gated. The same skills that find bugs can be used to exploit them. For the defender's view, read our breakdown of AI cybersecurity threats businesses need to watch.

Price, speed and availability

Standard pricing: $8 per million input tokens, $40 per million output tokens. Anthropic says that's lower than before.
Fast mode: up to 2.5x faster generation at $16 / $80 per million tokens. It's on Claude Code, the Claude Platform, Google Cloud and Microsoft Foundry. Mythos fast mode is only for verified organisations.
Speed: output up to 30% faster than the 5.1 models.
Context: 1M-token context window and up to 128k output tokens, according to Anthropic's developer docs.
Model IDs: claude-fable-5-2 and claude-mythos-5-2.
Launch perk: this week, eligible paid plans get one to three extra Fable "resets", and Fable can use up to 50% of weekly limits on Pro and standard Team seats.

Dev heads-up: the docs list four changes from Fable 5.1 that can break your code. One example: the older computer_20251124 computer-use tool now errors on every platform except Amazon Bedrock. Both models also need 30-day data retention unless Anthropic approves otherwise. Read the migration notes before you swap model IDs.

Safety: what Anthropic says it tested

The model was tested before release by external evaluators, including the UK AI Security Institute and METR. Anthropic rates the risk of high-stakes misalignment as low. In one deliberately weak sandbox test, Mythos 5.2 used credentials placed outside its sandbox in 3.1% of scenarios, compared with 21% for Mythos 5.1. Anthropic also admits the new models say they suspect they're being evaluated more often than earlier models, which makes testing harder.

Fable 5.2 also ships with what Anthropic calls its strongest-ever protection against distillation, where rivals train their models on Claude's outputs.

So… should you care?

Casual users: Fable is overkill for emails and essays. Anthropic itself says start with Opus 5.5. Our best AI tools for 2026 list is a better starting point.
Developers: if Opus 5.5 at high effort still isn't cutting it, Fable 5.2 is the upgrade path. Run your own evals first.
Security and biotech teams: apply to the verification programs if you need Mythos-level capabilities.

The AI coding race is moving fast. OpenAI is busy too; see our story on OpenAI acquiring Pierre for Codex.

FAQ

When was Claude Fable 5.2 released?

Anthropic announced Claude Fable 5.2 and Claude Mythos 5.2 on 9 November 2026. Fable 5.2 was available the same day.

What is the difference between Claude Fable 5.2 and Mythos 5.2?

They are the same underlying model. Fable 5.2 has strict cybersecurity and biology safeguards and is open to all users. Mythos 5.2 has more permissive safeguards and is limited to vetted organisations.

How much does Claude Fable 5.2 cost?

$8 per million input tokens and $40 per million output tokens. Fast mode costs $16 and $80.

Can I use Claude Mythos 5.2?

Only if your organisation is approved through Anthropic's Cyber Verification Program or Life Sciences Verification Program. Claude Enterprise customers can use it through Claude Security.