GitHub Adds Claude Sonnet 5.5 to Copilot for Paid Users

Developers can now choose Anthropic’s Claude Sonnet 5.5 in GitHub Copilot for everyday coding work, as GitHub rolls the option out across its paid plans. It is generally available, but that does not mean every eligible user will see it today. GitHub says the rollout is gradual, and organizations can control whether their teams get access.

The pitch: the same result with less work

GitHub describes Sonnet 5.5 as a fit for well-scoped jobs such as building features and fixing bugs. Its case for choosing the model rests on efficiency rather than a claim that it completes more coding tasks than Sonnet 5. In early testing, GitHub says Sonnet 5.5 matched Sonnet 5 on coding tasks while taking significantly fewer steps, using fewer tokens and tool calls, and finishing faster.

Those measures describe different parts of a coding job. Steps and tool calls capture how much work the model does along the way; tokens are units of input and output it processes. A faster finish could make a routine task less time-consuming. Fewer tokens could also affect its usage bill, but GitHub has not put a number on the savings in this announcement. Its comparison is an early test result, not a promise for every project.

Where the option appears

Copilot Pro, Pro+, Max, Business and Enterprise users are eligible. As the rollout reaches their accounts, they can select Sonnet 5.5 from the model picker rather than moving to a separate coding product. GitHub lists Visual Studio Code, Visual Studio, JetBrains IDEs, Xcode and Eclipse among the supported development environments.

The choice also extends beyond an editor. GitHub names Copilot CLI, its coding agent and Copilot app, github.com, and GitHub Mobile on iOS and Android. That gives developers a way to select the same model across several Copilot surfaces, whether they are working in an IDE or using one of GitHub’s other coding tools. Someone who cannot find it in the picker may simply be waiting for the staged rollout.

A team setting and a usage bill

For organizations, the default makes this more than an individual model choice. An administrator who wants to decide which models a team can use needs to check the existing policy or disable Sonnet 5.5 specifically. That setting is separate from rollout timing: a model may be permitted for a team yet not have appeared for all its eligible users.

GitHub says Sonnet 5.5 is billed at provider list pricing under usage-based billing. That makes its efficiency claim relevant to cost, but the early test result alone cannot tell a team what its coding work will cost. The practical test is whether the model finishes a team’s own well-scoped jobs with the results it needs, while using less time and fewer billable resources.