GitHub announced on September 28, 2026, that Anthropic’s Claude Sonnet 5.5 is generally available in GitHub Copilot. The change is less about a standalone model launch than about placing the model across coding agents, the Copilot app, CLI, IDEs, github.com, and GitHub Mobile so teams can select it inside existing development workflows.
GitHub says early testing showed Sonnet 5.5 matching Sonnet 5 on coding tasks while using fewer steps, tokens, and tool calls, and finishing noticeably faster. Those are early vendor test results; actual behavior will still depend on repository size, tool permissions, context, model settings, and task type.
The model is available to Copilot Pro, Pro+, Max, Business, and Enterprise users across Visual Studio Code, Visual Studio, Copilot CLI, Copilot coding agent, Copilot app, github.com, GitHub Mobile, JetBrains, Xcode, and Eclipse. GitHub notes that rollout is gradual, so an eligible account may not see the model immediately.
Enterprise and Business administrators can manage Sonnet 5.5 through the model policy in Copilot settings. With default model enablement on, new models are enabled automatically; administrators who disable the global default or explicitly disable this model can prevent access. Model supply and model policy therefore become one governance task rather than a purely individual choice.
GitHub says Sonnet 5.5 is billed at provider list pricing under usage-based billing. Fewer tool calls do not automatically mean a lower total bill: an agent may attempt longer tasks, touch more files, or use a higher effort setting. Teams should measure tokens, tool calls, wall time, retries, and human review together.
Putting Sonnet 5.5 inside Copilot shortens the distance from a model release to a working development process. A model choice can follow a developer between the IDE, CLI, cloud agent, and mobile surfaces, but that consistency also makes permissions, retention, branch scope, repository scope, and approval policy more important.
A safer adoption path is to start with a low-risk repository with explicit tests and review, record completion, rollback, tool-call, and cost data, and then decide whether to expand to production code or long-running cloud agents. General availability means the model is in the product’s supported supply; it does not remove the need for team-level validation.
