Summary
Anthropic announced Claude Sonnet 5.5 on September 28, 2026, reporting higher scores on several evaluations, output generation more than 30% faster than Sonnet 5, and task costs up to 30% lower in its testing. The model is available across Anthropic’s platforms and major cloud services.
Anthropic announced Claude Sonnet 5.5 on September 28, 2026, describing it as a faster, lower-cost successor to Sonnet 5. The company says the model generates output more than 30% faster and costs up to 30% less per task in its testing. It is intended for everyday, well-scoped work such as fixing bugs and producing documents, slides and spreadsheets; Anthropic positions Opus 5.5 as the stronger choice for complex work requiring sustained judgment.
Contents
Coding and knowledge-work results
Anthropic reports a score of 70.6% for Sonnet 5.5 on Terminal-Bench 4.0, an evaluation of multi-step coding tasks in a command-line environment. Sonnet 5 scored 10.3% on the same benchmark. On GDPval-AA, which tests work across occupations, Sonnet 5.5 scored two points below Opus 5.5. The company also reports gains in computer use, chart recognition and long-horizon knowledge work.
These results are evaluation scores, not a single measure of general capability. Anthropic says Opus 5.5 remains stronger at complex, open-ended work that demands sustained judgment. It also notes that Artificial Analysis ran GDPval-AA and AA-Briefcase on a pre-release version with a bug that could affect requests using structured outputs. Anthropic says the issue has been fixed and expects any effect on those scores to be small.
Speed and pricing
Sonnet 5.5 has the same listed token prices as Sonnet 5: $2 per million input tokens, $10 per million output tokens, $0.20 per million cache-read tokens and $2.50 per million cache-write tokens. Anthropic attributes its lower cost per task to using fewer tokens to complete work, and says its tests found savings of up to 30% compared with Sonnet 5. That is a task-cost result reported by the company, rather than a reduction in the listed price per token.
Users can adjust the model’s effort setting to trade speed and token use against more extensive work. Anthropic says the default is Medium in its apps and Claude Code, and High on the Claude Platform.
Safety and availability
Anthropic says Sonnet 5.5 improved on or matched Sonnet 5 on most measures in an automated behavioral audit covering roughly 1,850 scenarios. Because the company says its cybersecurity capabilities have improved substantially over Sonnet 5, the new model launches with safeguards similar to those on Opus 5.5. Higher-risk cybersecurity tasks will fall back to Sonnet 5, while routine software development remains available. Sonnet 5.5 uses the same biology safeguards as Sonnet 5; Anthropic says most research, education and clinical work is unaffected.
The model is available on the Claude Platform and across Anthropic’s listed platforms, including Amazon Web Services, Google Cloud and Microsoft Azure. Developers can access it using the model identifier claude-sonnet-5-5. Anthropic also says it is available with zero data retention.