Claude Opus 4.7 Is Here and It Changes the Coding Model Race

Anthropic dropped Claude Opus 4.7 with no rollout or waitlist, just live across Claude Code, the API, and every major cloud provider. It scores 87.6% on SWE-bench Verified, ships a new xhigh effort level that sits between high and max, and delivers 3x sharper vision for reading screenshots, diagrams, and PDFs. Agentic coding is noticeably better on long multi-step tasks, and pricing stays the same as 4.6. If you're already on 4.6, the upgrade is a no-brainer, swap the model ID and go. If you're on Sonnet for cost reasons, 4.7 doesn't change that calculus. Below: real benchmark numbers, what xhigh actually does, where the vision improvements show up, and the edge cases where 4.6 still wins.