Claude Opus 5.5 arrives — and becomes the default in Claude Code
Anthropic's new frontier model tops the major intelligence indexes at a lower price than Opus 5, and it is now the default in Claude Code on Max — with design, slides, and docs tooling arriving alongside it.
Anthropic has shipped Claude Opus 5.5, and the company is framing it as a clean step up rather than a side grade: it tops the Artificial Analysis Intelligence Index v4.3 and the Vals Index, undercuts Opus 5 on price, and skips the 30-day retention terms that accompanied Fable-style deployments elsewhere. Sonnet and Haiku 5.5 are said to follow within weeks, which would refresh the whole Claude lineup at once.
The more consequential move for working developers is the default flip: Opus 5.5 is now the default model in Claude Code on Max. Anthropic pitches it at roughly Fable 5.1 quality on most tasks, about 30% faster and 40% cheaper per task than Opus 5, with writing that follows style instructions more reliably — the trait that matters most for agentic coding sessions where the model both edits files and drafts the commit messages around them.
Alongside the model, Claude Code is growing beyond code. A desktop and web beta adds Claude Design, Slides, and Docs: Claude can now produce UI mockups, review decks, and draft RFCs from a repository in-session, with live editing in a preview pane. The through-line is a coding agent that treats the repo as context for every artifact a team produces — not just the diff.