According to CoinMeta, Anthropic has upgraded the underlying scanning model of Claude Security to Mythos v5 and made it available for public testing to all Claude Enterprise customers. After enterprises integrate with the GitHub repository, Mythos v5 will scan for code vulnerabilities and provide CWE classification, confidence level, severity, and repair recommendations. The biggest change this time is that the previously strictly restricted Mythos v5 has been incorporated into the existing security products. Enterprises still cannot directly invoke Mythos v5; the model only runs in the background during scanning, and users receive vulnerability reports and repair suggestions. Subsequently, when modifying code within Claude Code, the existing enterprise models will still be used, and patches must be approved manually. Scanning is charged according to the current packages; there is no need to purchase Mythos v5 separately. Anthropic also plans to integrate Mythos v5 into partners' security products and has launched a $35 million Claude fund to assist open-source projects in identifying and fixing vulnerabilities.