Claude Sonnet 5.5 is now generally available in GitHub Copilot, giving eligible developers another model option for everyday coding work. GitHub positions it for well-scoped feature development and bug fixing, while reporting that its early testing matched Claude Sonnet 5 on coding tasks with fewer steps, tokens, and tool calls.
The model is available across Copilot Pro, Pro+, Max, Business, and Enterprise plans. However, the rollout is gradual, so an eligible user may not see Claude Sonnet 5.5 immediately. GitHub has not supplied a date for when the rollout will be complete.
What changed in GitHub Copilot
GitHub announced Claude Sonnet 5.5 on September 28, 2026. The model can be selected from the model picker on supported Copilot surfaces, subject to plan eligibility and rollout status. The announcement covers a broad set of clients and services:
- Visual Studio Code and Visual Studio
- Copilot CLI
- GitHub Copilot coding agent
- GitHub Copilot app and github.com
- GitHub Mobile on iOS and Android
- JetBrains IDEs, Xcode, and Eclipse
The supported-models documentation also lists Claude Sonnet 5.5 in its client-availability table. That confirms broad client coverage, but it does not establish that every surface receives the model at the same time or offers identical capabilities.
Plans, access, and administration
Claude Sonnet 5.5 is listed for Copilot Pro, Pro+, Max, Business, and Enterprise. Individual subscribers can check the model picker on a supported surface. For Business and Enterprise deployments, administrators can manage access through Copilot model policies.
GitHub says new models are enabled by default unless an administrator has disabled the global default or the specific model. Organizations that need tighter control should therefore review their model-policy settings rather than assume that every newly available model is blocked until manually approved.
The available documentation does not describe a Claude Sonnet 5.5-specific migration procedure, API configuration, or troubleshooting workflow. There are also no documented region-specific prerequisites in the supplied material.
What the performance claim means
GitHub reports that, during its early testing, Claude Sonnet 5.5 matched Claude Sonnet 5 on coding tasks while using significantly fewer steps, tokens, and tool calls and completing tasks faster. This is a vendor-reported test result, not an independent benchmark or a guarantee for every repository and workload.
For developers, the practical takeaway is that Sonnet 5.5 may be worth evaluating for routine, well-defined changes where interaction efficiency matters. Teams should still compare results against their own coding tasks, review generated changes, and account for the cost of model interactions before standardizing on it.
Billing and AI-credit allowances
Copilot usage is token-based. Input tokens, output tokens, and cached tokens contribute to usage, and the resulting cost depends on the selected model and the amount of interaction. The supplied Copilot pricing documentation lists Claude Sonnet 5.5 at:
- $2.00 per 1 million input tokens
- $0.20 per 1 million cached input tokens
- $2.50 per 1 million cache-write tokens
- $10.00 per 1 million output tokens
One GitHub AI credit equals $0.01 USD. The documented monthly allowances are:
- Copilot Pro: 1,500 AI credits
- Copilot Pro+: 7,000 AI credits
- Copilot Max: 20,000 AI credits
- Copilot Business: 1,900 AI credits per user
- Copilot Enterprise: 3,900 AI credits per user
Business and Enterprise credits contribute to a shared organization or enterprise pool. These allowances provide budget context, but they do not translate into a fixed number of Sonnet 5.5 tasks because actual consumption varies with prompt size, output size, caching, and tool interactions.
When included credits are exhausted, individual users can upgrade, purchase additional usage where permitted by budgets and caps, or wait for the next monthly reset. Business and Enterprise users can request additional budget from an administrator. Code completions and next edit suggestions are not billed in AI credits and remain unlimited for paid Copilot plans under the documented counting mechanism.
Client version guidance
For Visual Studio 2026, the supported-models page lists version 17.14.6 as the minimum listed version for Claude Sonnet 5.5. The same model-specific table does not list a minimum version for Visual Studio Code, JetBrains IDEs, Xcode, or Eclipse; that absence should not be interpreted as confirmation that every older version is supported.
Separately, GitHub’s general usage-based billing guidance recommends current minimum client versions for accurate usage information: VS Code 1.120, Visual Studio 2022 17.14.33, Visual Studio 2026 18.6.0, JetBrains plugin 1.9.1, Eclipse plugin 0.18.0, Xcode extension 0.50.0, and Copilot CLI 1.0.48. These are billing-display recommendations, not universal Claude Sonnet 5.5 compatibility requirements.
What you should do
- Check the model picker. Look for Claude Sonnet 5.5 in the Copilot surface you use, while allowing for the gradual rollout.
- Update relevant clients. Visual Studio 2026 users should meet the model-specific 17.14.6 minimum listed by GitHub. Other users should review the general billing-client recommendations if accurate usage information is important.
- Review administrative policies. Business and Enterprise administrators should verify global default and model-specific policy settings before deploying the model broadly.
- Estimate usage before switching workflows. Use the published token rates and plan allowances as a starting point, but measure actual consumption on representative coding tasks.
- Evaluate the reported efficiency claim locally. GitHub’s early-testing results may not predict performance, speed, or token use for every codebase.
Availability caveats
Claude Sonnet 5.5 is described as generally available, but availability is still rolling out gradually. The supplied sources do not provide a completion date, regional requirements, or model-specific usage limits. They also do not establish simultaneous availability or identical behavior across all listed clients.
The documentation supplied for this announcement also does not describe Sonnet 5.5 as a security feature or provide model-specific data-handling guarantees. General Copilot content-filtering and public-code-matching behavior documented for default models should not be treated as a Sonnet 5.5-specific security claim.
