Opus 4.7 Can See Better: And More About Anthropic’s Concession Prize

Anthropic released Opus 4.7, with early users reporting stronger image understanding, coding, and reasoning than Opus 4.6.

The update highlights improvements across Claude’s core capabilities, including better performance on tasks like SWE-bench.

At the same time, Anthropic is tightening cybersecurity controls with a Cyber Verification Program and a use-case form.

Those safeguards appear designed to support a broader rollout of the more advanced Mythos model later on.