Anthropic Launches Claude Fable 5.2 and Claude Mythos 5.2 AI Models for Technical Work

Anthropic Launches Claude Fable 5.2 and Claude Mythos 5.2 AI Models for Technical Work

Anthropic says its two new models are built for demanding technical tasks including coding, reasoning, science and cybersecurity.

Two new artificial intelligence models from Anthropic landed with a pointed message. “Our most capable models for daunting technical work”, said the summary on Anthropic’s news page, and it tells you exactly who Anthropic is pitching to. Not casual users asking chatbots to write birthday cards. Engineers. Researchers. Cybersecurity teams. People who need an AI that can hold its nerve on a hard problem for a long time.

Anthropic posted the announcement on 9 November 2026, confirming that Claude Fable 5.2 and Claude Mythos 5.2 were ready to use.

What Are Claude Fable 5.2 and Claude Mythos 5.2?

The two models share the same underlying technology but serve different audiences.

Claude Fable 5.2 is the general-purpose entry point. It’s built for coding, finance, science and what it describes as long-horizon agentic work, meaning tasks where an AI doesn’t just answer a question but works through a multi-step problem over an extended period, making decisions along the way. Think of it less like a calculator and more like a colleague who can be left to get on with something complicated and come back with results.

Claude Mythos 5.2 is the restricted sibling. It uses the same core model as Fable 5.2 but is available only to verified organisations. Anthropic says it’s aimed specifically at cybersecurity engineering and life-sciences research. The company was direct about the intended use: “Claude Mythos 5.2 is our most capable model for cybersecurity engineering.” Getting access means going through Anthropic’s verification and trusted-access programmes. It’s not available to anyone who simply signs up.

Speed and Pricing

Speed is one of the headline claims. Anthropic says both models generate outputs up to 30% faster than their 5.1 predecessors, Claude Fable 5.1 and Claude Mythos 5.1. There’s also an optional fast mode that can push token-generation speed up to 2.5 times higher than standard operation. These are company-reported figures and have not yet been independently benchmarked.

On pricing, Anthropic has moved in the direction most developers will welcome. The stated rate is $8 per million input tokens and $40 per million output tokens. At current exchange rates that works out to around £6.30 per million input tokens and around £31.50 per million output tokens, though GBP pricing has not been independently confirmed and businesses should verify directly with Anthropic or their cloud provider before budgeting.

Claude Fable 5.2 is available through Anthropic’s own platform and through selected cloud marketplaces: Amazon Web Services, Google Cloud and Microsoft Foundry. Claude Mythos 5.2 remains behind Anthropic’s access controls.

Safety Testing

Anthropic says it carried out extensive pre-release safety testing before releasing either model. The company reports running checks across around 1,900 behavioural-audit scenarios, a testing process designed to identify how the models behave when pushed towards problematic outputs. The restricted access structure around Mythos 5.2 is itself part of the safety picture: by limiting it to vetted organisations, Anthropic is attempting to control who can use the more capable cybersecurity-focused version.

But those assurances come from Anthropic itself. The safety claims, the speed improvements, the performance benchmarks: all of them are self-reported at this stage. Independent researchers and enterprise buyers will want to test the models against their own requirements before drawing firm conclusions.

That’s a reasonable caution. It doesn’t mean the claims are wrong. It means they haven’t been verified yet by anyone outside the company.

The Broader Picture

Anthropic is one of the main players in the current generation of large language model development, alongside OpenAI, Google DeepMind and Meta. The release of Fable 5.2 and Mythos 5.2 continues a pattern across the industry: models are getting faster, prices are coming down, and the focus is shifting from general-purpose assistants towards tools built for specific professional domains.

The Mythos line in particular reflects a broader question the industry is working through: how do you make a genuinely capable AI available to the people who need it for legitimate security and research purposes, without making it equally available to those who’d use it badly? Anthropic’s answer, for now, is a verification gate. Whether that holds up as the models spread across cloud marketplaces nobody knows yet.

On top of that, the announcement said it plainly: “Claude Fable 5.2 and Claude Mythos 5.2 are available now.”

What This Means for Kent Residents

For businesses, NHS organisations and councils in Kent considering advanced AI tools, these models are now accessible through major cloud platforms, though any use by public bodies such as NHS Kent and Medway ICB, Kent County Council or Medway Council would need to go through proper procurement, data protection and UK public-sector AI guidance before anything is deployed. For developers and technical teams in the county, Anthropic says the new models are available at a lower price, though exact GBP costs should be confirmed directly with the relevant cloud provider.

Source: @AnthropicAI

⚡

Anthropic Launches Claude Fable 5.2 and Claude Mythos 5.2 AI Models for Technical Work Quiz

5 questions