Claude Sonnet 5.5 is Anthropic's mid-tier foundation model in the Claude 5.5 family, built for agentic terminal coding, visual comprehension, and document generation with adjustable reasoning effort levels.
What the product does and how it is positioned
Claude Sonnet 5.5 is positioned as an intermediate model within the Claude 5.5 family, serving as a lower-latency complement to Claude Opus 5.5. It focuses on well-scoped everyday tasks, multi-file code editing, bug fixing, and the generation of structured documents, spreadsheets, and presentation slide decks.
The model supports configurable reasoning effort settings across development environments, defaulting to Medium in Claude apps and Claude Code and High on the Claude Platform. It also incorporates safety mechanisms including sandbox containment protections, account-bound preserved thinking, and domain-specific safeguards for cybersecurity and biological queries.
Source-supported ways to use the product
The official customer story from Epic Games reports using the model for system design audits, data flow reviews, and multi-hour development sessions managing gameplay system architecture code.
The company presents a customer use case with Zendesk, which reports feeding the model real customer support inquiries and escalation requests to accelerate resolution times.
The official customer story from Box reports using the model to recheck data across sensitive source documents, identifying errors in financial services and healthcare contexts.
Claude Sonnet 5.5 allows users and developers to calibrate reasoning depth through adjustable effort parameters. Lower effort levels prioritize faster generation and reduced token counts for routine programming or conversational tasks, while higher effort tiers enable extended reasoning loops for rigorous verification.
Different deployment surfaces apply distinct standard settings. Claude Code and end-user Claude applications default to Medium effort, whereas the Claude Platform defaults to High effort. When migrating legacy configurations that operated with thinking turned off, developers are required to configure the between_tools parameter to keep upfront thinking inactive.
Checks to run with your own material and workflow
What was checked and when
Answers based on the source-checked product record
Claude Sonnet 5.5 defaults to Medium effort in Claude apps and Claude Code, while the Claude Platform defaults to High effort.
Routine software development and bug fixing run normally, while higher-risk cybersecurity tasks automatically fall back to Sonnet 5 unless approved via the Cyber Verification Program.
Developers who run Sonnet with thinking turned off must switch to the new between_tools setting to keep upfront thinking deactivated before transitioning to Sonnet 5.5.
Claude Sonnet 5.5 is available on Amazon Web Services, Google Cloud, Microsoft Azure, the Claude Platform, Claude Code, and Claude applications.
The model supports zero data retention configurations, employs anti-distillation classifiers, restricts reasoning extraction through preserved thinking, and operates within evaluated container boundaries.
