Anthropic announced the release of Claude Sonnet 5.5 on September 28, 2026, positioning it as a more efficient and cost-effective option for everyday enterprise tasks. The new model is the second in the Claude 5.5 family, following the flagship Opus 5.5. While the API pricing for Sonnet 5.5 remains consistent with Sonnet 5 at $2 per million input tokens and $10 per million output tokens, Anthropic indicates that the actual cost per task can decrease by as much as 30%. This reduction stems from Sonnet 5.5's ability to complete tasks using fewer tokens and making fewer tool calls.
The company reports that Sonnet 5.5 generates outputs over 30% faster than Sonnet 5, making it the fastest Sonnet model to date. This speed improvement is particularly noticeable in coding, where Sonnet 5.5 scored 70.6% on the Terminal-Bench 4.0 agentic coding evaluation, a substantial increase from Sonnet 5's 10.3%. On the CursorBench 4.0, which simulates real coding sessions, Sonnet 5.5 achieved 55.5%, placing it closely behind Opus 5.5's 57.8%. Anthropic also highlighted that on the FrontierCode benchmark, Sonnet 5.5 at a "High" effort setting scored 10 points higher than Sonnet 5 at approximately one-fifteenth the cost per task.
Beyond coding, Sonnet 5.5 demonstrates improved performance in knowledge-work benchmarks. On GDPval-AA, a test covering various occupations, Sonnet 5.5 scored 1,844 points, nearly matching Opus 5.5's 1,846 points and significantly surpassing Sonnet 5's 1,449 points. Anthropic states that Sonnet 5.5 is well-suited for structured, everyday tasks such as bug fixing, document creation, and spreadsheet generation, while Opus 5.5 remains designed for more complex work requiring extensive judgment.
The release of Sonnet 5.5 also includes enhanced cybersecurity safeguards, which Anthropic previously reserved for its higher-capability models. This marks the first Sonnet model to incorporate these frontier-style cyber limits and classifiers designed to prevent reasoning extraction. While routine software development and vulnerability remediation are expected to function normally, certain higher-risk cybersecurity requests will revert to Sonnet 5. Anthropic plans to expand its Cyber Verification Program to allow approved defenders access to more advanced capabilities in Sonnet 5.5, Opus 5.5, and its Mythos models.
Sonnet 5.5 is available across major cloud platforms, including Amazon Web Services, Google Cloud, and Microsoft Azure. Anthropic also confirmed that Claude Haiku 5.5, intended for high-volume and cost-sensitive applications, will join the Claude 5.5 family in the coming weeks.