Claude Opus 5 — the new everyday Opus for coding and long-running agents

Anthropic released Claude Opus 5 as a more efficient model for difficult coding and professional work.

What changed

• Better software-engineering performance, root-cause analysis, self-verification, and long-horizon execution.
• The same base API price as Opus 4.8: $5 per million input tokens and $25 per million output tokens.
• Availability on all Claude platforms, including Claude Code, where it became the default Opus model.
• Effort settings for trading speed and token use against maximum capability, plus a faster mode at a higher price.
• New API beta features for changing tools mid-conversation and automatically routing requests when safeguards intervene.

Why it matters

Opus 5 is positioned for everyday high-difficulty work rather than only exceptional tasks. Its stronger verification is useful evidence, not a substitute for running the project's own tests and reviewing changes before merge.

Official announcement: Introducing Claude Opus 5 · Claude Code 2.1.219 release notes