Claude Opus 5 is now available in the model selector, giving you Anthropic's latest flagship model for your most demanding threads. It comes in two variants: the standard model for fast, high-quality responses, and Claude Opus 5 (Extended Thinking), which automatically applies extended reasoning when a prompt calls for deeper step-by-step analysis.
How to use it
Open the model dropdown at the top of the message composer.
Select Claude Opus 5 for everyday high-quality responses, or Claude Opus 5 (Extended Thinking) when you want the model to reason through complex, multi-step problems before answering.
Send your message as usual — the extended thinking variant will decide on its own how much reasoning a given prompt needs, so there's nothing extra to configure.
Tips
Claude Opus 5 costs more per message than Claude Sonnet 5 — roughly 1.7x — so it's best suited to tasks where the extra quality or reasoning depth is worth it, rather than routine drafting.
If you previously used Claude Opus 4.8, you'll now find it listed under legacy models in the selector. We recommend switching to Claude Opus 5 for new threads, since it's the model our automatic routing now uses for top-tier requests.
Existing threads that were already using Claude Opus 4.8 will keep working, but new threads default to Claude Opus 5 when a top-tier model is selected automatically.