Claude Fable 5 and Claude Mythos 5: Anthropic’s Latest AI Models

Anthropic has announced the launch of Claude Fable 5, a new Mythos-class AI model designed for general use with added safety protections. The company says Fable 5 is its most capable publicly available model yet, delivering state-of-the-art performance across software engineering, knowledge work, vision, scientific research, and other demanding tasks. According to the company, the model performs especially well on long, complex assignments, where its advantage over earlier Claude systems becomes more pronounced.

Alongside Fable 5, Anthropic is introducing Claude Mythos 5 for a limited group of cyber defenders and infrastructure providers. Mythos 5 uses the same underlying model as Fable 5 but removes some safeguards in controlled settings. It will first be deployed through Project Glasswing in collaboration with the US government, as an upgrade to Claude Mythos Preview, and the company says it plans to expand access later through a broader trusted access program. Anthropic says Mythos 5 has the strongest cybersecurity capabilities of any model it has tested.

The company says the release reflects both the opportunities and risks of more advanced AI. To reduce potential misuse, Fable 5 includes new safety classifiers that detect risky prompts related to cybersecurity, biology and chemistry, and model distillation attempts. When these classifiers are triggered, the request is redirected to Claude Opus 4.8 instead of Fable 5. Anthropic says the safeguards are intentionally conservative at launch, which means some harmless requests may be flagged, but it reports that the fallback happens in less than 5% of sessions on average.

Anthropic says the model has shown strong results in software engineering. In early testing, partners such as Stripe reportedly used Fable 5 to compress months of engineering work into days, including a large-scale migration across a 50-million-line Ruby codebase. The company also says the model achieved top results on coding evaluations that measure production-quality output and difficult real-world development tasks.

In knowledge work, Fable 5 is described as especially strong in finance, analytics, document reasoning, and chart interpretation. The company says it outperformed previous models on several internal and partner benchmarks. In vision, it can extract precise data from scientific figures and handle complex screenshot-based tasks, including reconstructing web app source code from images. Anthropic also says the model shows major gains in memory and long-context work, maintaining focus over millions of tokens.

Mythos 5 has also been tested in scientific research. Anthropic says its internal protein-design experts used it to accelerate parts of drug discovery, and that it helped generate novel hypotheses in molecular biology and genomics. The company says these results point to meaningful future applications in life sciences and therapeutic research.

Anthropic is pricing both models at $10 per million input tokens and $50 per million output tokens. Claude Fable 5 is available now, while Claude Mythos 5 remains limited to approved partners and select researchers under trusted access programs.