Anthropic is launching a public version of its Mythos AI model, but with implemented safeguards that prevent its use in high-risk areas such as cybersecurity. This model had previously garnered global attention during an earlier preview this year for its ability to identify software vulnerabilities.
The startup announced on Tuesday that the new Claude Fable 5 is its most powerful model to date, designed for broader application, and praised its performance in software engineering and data analysis.
Anthropic had previously stated that Mythos had discovered thousands of software vulnerabilities. Following this, the company restricted access to the model to approximately 200 organizations, including the U.S. government, through its "Glasswing" program.
As the two giants of the AI industry race to go public, this company, valued at $965 billion, could potentially extend its lead in valuation over competitor OpenAI if it makes its technological capabilities more widely available.
The company stated it has conducted extensive testing to ensure users cannot manipulate the new model to bypass its guidelines and perform restricted operations.
"Suppose I am a university student and ask the model to 'help me find network vulnerabilities in X software package or code.' The model will refuse the request, and Fable 5 will fall back to Opus 4.8 to provide a response," said Dianne Penn, Head of Product Management, Research, and Labs at Anthropic.
Penn noted that based on early customer feedback, while Fable 5 is more expensive, it requires fewer tokens to execute tasks, thereby reducing the overall cost per task.
Anthropic also stated that users with access to the Claude Mythos preview version, which lacks security safeguards, will be able to upgrade to the new Claude Mythos 5.
The company indicated it plans to gradually expand access through a more "systematic trusted access program."
The company announced that pricing for both models is set at $10 per million input tokens and $50 per million output tokens.
Comments