Anthropic rolled out a safer version of Mythos AI model for public use, called Claude Fable 5.
The AI startup said the Fable 5 release was made possible after it added safeguards to block answers related to high-risk topics such as cyberattacks and dangerous biological materials.
Risky requests are automatically filtered, and the system switches to the Claude Opus 4.8 model to deliver safe answers.
The model is now available to enterprise customers and paid subscribers.
Led by CEO Dario Amodei, Anthropic gave limited access to Mythos earlier this year because of concerns that it could be misused. It says the new model delivers major improvements in software engineering and complex knowledge tasks, outperforming its earlier Claude Opus 4.8 on several benchmarks.
The launch also comes as the company prepares for a widely expected IPO.