16 April 2026claude#model#opus#coding#vision
Claude Opus 4.7 — harder software tasks, stronger vision, and better verification
Opus 4.7 improved difficult software engineering, long-running work, visual understanding, instruction following, and output verification.
Anthropic released Claude Opus 4.7 for complex software engineering and professional work.
What changed
- Stronger performance on difficult coding tasks that require sustained reasoning.
- More consistent instruction following and greater willingness to verify work before declaring success.
- Higher-resolution image understanding and better visual output for interfaces, slides, and documents.
- Claude Code support for
xhigheffort and eligible auto-mode workflows arrived with the model.
Why it matters
Opus 4.7 was aimed at tasks users previously needed to supervise closely. Verification behavior helps, but passing tests and reviewing the resulting diff remain the reliable completion standard.