Anthropic releases Claude Opus 4.7 with reduced cyber capabilities

Anthropic launched Claude Opus 4.7, an improvement over prior models in software engineering and high-resolution image inspection, but with deliberately reduced cybersecurity abilities compared to the unreleased Mythos Preview. It is positioned as less broadly capable than Mythos for safety reasons. It matters because it shows Anthropic’s strategy of releasing safer, specialised variants alongside controlled frontier models.