Claude Sonnet 5.5 is now available in Kiro

Claude Sonnet 5.5 is available today across Kiro IDE, CLI, Crew, and Web. It is Anthropic’s fastest Sonnet model to date and a clear upgrade over Claude Sonnet 5. In our internal benchmarking, Sonnet 5.5 performed better than Sonnet 5, while being more efficient in both token usage and speed. Sonnet 5.5 is the second model in the Claude 5.5 family, following Opus 5.5.

As per Anthropic data, Sonnet 5.5 generates output 30%+ faster than Sonnet 5, which makes it Anthropic’s fastest Sonnet yet. It’s also smarter. It scores 70.6% on Terminal-Bench 4.0 for agentic coding, up from 10.3% on Sonnet 5, and nearly matches Opus 5.5 on GDPval-AA, a benchmark for real-world knowledge work. Like Opus 5.5, it writes more clearly than earlier models, and early testers described it as a better partner for collaboration than Sonnet 5.

Sonnet 5.5 is built for well-scoped coding and everyday work of building features, fixing bugs, and reviewing code. It usually checks its own work before it calls a change done. It also does well in agent workflows. In early testers’ head-to-head runs, it finished multi-step tool work in fewer steps than Sonnet 5, so it works well as a faster, lower-cost subagent alongside Opus 5.5.

Sonnet 5.5 ships with Adaptive Thinking and you can steer depth and cost with the effort setting, using five levels from low to max. The model also has a 1M token context window and up to 128K output tokens, with no extra charge for long context.

Because its cybersecurity capabilities are comparable to Anthropic’s most capable models of the Opus series, Sonnet 5.5 also comes with cyber safeguards and fallbacks developed for them. While most routine software development including finding and fixing bugs in your code works as usual, some higher-risk requests may be blocked. Those tasks can be rerun on Sonnet 5. The model’s biology safeguards are the same as those of Sonnet 5, so most life sciences work is unaffected.

Claude Sonnet 5.5 is gradually rolling out with experimental support to Kiro Pro, Pro+, Pro Max, and Power customers in us-east-1 (IAD) and eu-central-1 (FRA). It ships with a full 1M context window and a 1.3x credit multiplier.