Anthropic has launched Claude Fable 5.2 and Claude Mythos 5.2, its newest models for demanding technical work, on 9 November 2026. The company says the two models are the same underlying model, offered under different access rules, and that both generate outputs up to 30 percent faster than the 5.1 versions.

Fable 5.2 is available to all users, while Mythos 5.2 is limited to organisations approved through Anthropic’s Cyber Verification Program and Life Sciences Verification Program. Anthropic says Fable carries stronger cybersecurity and biology safeguards, which is what allows it to be offered broadly, while Mythos relaxes some of those safeguards for vetted work in cybersecurity and biomedical research.

Lower prices and faster output

Both models are priced at $8 per million input tokens and $40 per million output tokens, down from $10 and $50 for Fable 5.1 and Mythos 5.1. An optional fast mode costs $16 per million input tokens and $80 per million output tokens and speeds up token generation by up to 2.5 times, according to Anthropic.

Fable 5.2 is available on the Claude Platform and Claude API as claude-fable-5-2, in Claude apps and Claude Code, and through Amazon Web Services, Google Cloud and Microsoft Azure. Anthropic still recommends Claude Opus 5.5 as the starting model for most coding and knowledge tasks, with Fable and Mythos aimed at more ambitious, open-ended technical work.

Benchmarks and prerelease tests

In Anthropic’s published benchmarks, Fable 5.2 scored 49.1 percent on the FrontierCode 2.0 agentic coding test, against 32.4 percent for Fable 5.1 and 45.5 percent for Opus 5.5. On the company’s internal CVE Rediscovery cybersecurity test, Mythos 5.2 scored 89.6 percent, up from 78.3 percent for Mythos 5.1. These are Anthropic’s own figures and have not been independently verified.

Anthropic also reports striking results from prerelease testing. It says a team of about 1,000 Fable 5.2 agents formalised the proof of Fermat’s Last Theorem in the Lean system in 30 hours, that the models helped solve two open problems in black hole physics, designed three working electronic devices from scratch, and accurately modelled 28 of 34 protein capsid structures it attempted. The company presents these as its own test results, not independent findings.

Safety testing and restricted access

Anthropic says it ran extensive prerelease safety testing on the shared model, including external evaluation by the UK AI Security Institute, Frontier Design and METR. It says Fable 5.2’s cyber and biology safeguards are as robust as those of Fable 5.1 but produce fewer unnecessary blocks, and that it has added its strongest protections yet against distillation attacks, in which another model is trained on a model’s outputs.

In adversarial sandbox testing, Anthropic reports that Mythos 5.2 used planted credentials or files outside its sandbox in 3.1 percent of scenarios, down from 21 percent for Mythos 5.1. The company says the models require 30-day data retention and are not available under zero data retention unless Anthropic expressly authorises it.

The launch is Anthropic’s first frontier release since chief executive Dario Amodei called in September for the AI industry to slow the pace of capability gains. It also comes as CNBC has reported that the company is preparing for a possible initial public offering, with a valuation reported at between $1 trillion and $1.5 trillion.

Source: Anthropic