If Opus is the specialist you call for the hard judgment calls, Claude Sonnet 5.5 is the teammate who clears the everyday desk—and this week Anthropic says that teammate just leveled up.
Released September 28 as the second model in the Claude 5.5 family, Sonnet 5.5 is pitched as a faster, lower-cost complement to Claude Opus 5.5: strong at well-scoped work like bug fixes, polished docs, slides, and spreadsheets, with a sharper eye for design. Anthropic says it generates outputs more than 30% faster than Sonnet 5 and, because it typically needs fewer tokens for the same job, costs up to 30% less per task—while list pricing stays at Sonnet 5’s $2 per million input tokens and $10 per million output tokens.
The benchmark jumps are hard to miss. On Terminal-Bench 4.0, an agentic coding test, Sonnet 5.5 scores about 70.6%, up from Sonnet 5’s roughly 10.3%. On GDPval-AA, a knowledge-work suite across many occupations, Sonnet 5.5 lands within about two points of Opus 5.5 (1,844 versus 1,846). Early testers called out by Anthropic include Epic Games, SpaceXAI, Slack, Zendesk, and Unity—citing snappier coding, fewer wasted steps or tool calls, and stronger polish on real workflows.
Safety-wise, Sonnet 5.5 is the first Sonnet to ship with cyber safeguards and fallbacks akin to Anthropic’s most capable models, plus distillation protections meant to block industrial-scale extraction of the model’s capabilities. It is available across Anthropic’s platforms and major clouds, with Haiku 5.5 still queued for the coming weeks.
Why it matters
Most of us do not need a “max brain” on every email, ticket, or spreadsheet. A mid-tier model that nearly matches the flagship on practical work—but answers faster and burns fewer tokens—is the difference between AI that feels like a demo and AI that fits a Tuesday afternoon. Better default tools at fairer per-task cost is how assistants quietly get useful for the rest of us.
What’s next
Developers can call `claude-sonnet-5-5` now; watch how teams shift everyday coding and support loads onto Sonnet while saving Opus for the gnarliest judgment calls—and how Haiku 5.5 fills the high-volume lane when it arrives.
Sources: Anthropic — Introducing Claude Sonnet 5.5, THE DECODER — Sonnet 5.5 nearly matches Opus 5.5