Anthropic launches Claude Sonnet 5.5 with faster coding and Copilot access
Claude Sonnet 5.5 is here for everyday coding and knowledge work. Anthropic reports faster runs and lower cost per task, while GitHub is rolling it into Copilot across its paid plans.
Anthropic launched Claude Sonnet 5.5 on September 28, positioning it for the coding and knowledge-work tasks that do not need its more expensive Opus tier. GitHub has separately made the model available in Copilot. That gives developers a practical choice: keep a familiar assistant surface, but test whether a faster model can finish routine work with fewer steps.
Faster work without a cheaper token price
Anthropic says Sonnet 5.5 generates output more than 30% faster than Sonnet 5 and costs up to 30% less per task in its testing. The per-token API prices have not fallen: input remains $2 per million tokens, output $10, and cache reads $0.20. The claimed saving comes from needing fewer tokens to complete the same work, not from a new discount on every token.
The company points to stronger agentic coding, document creation, slide work, and image understanding. GitHub says its early Copilot tests also saw fewer steps, tokens, and tool calls for well-scoped feature and bug-fix tasks. These are provider-reported results, not a guarantee for a particular repository. Anthropic still describes Opus 5.5 as stronger for complex, open-ended work requiring sustained judgment.
Where builders can use Sonnet 5.5
Developers can select claude-sonnet-5-5 on the Claude Platform. Anthropic also lists availability through AWS, Google Cloud, and Microsoft Azure. GitHub says the model is available to Copilot Pro, Pro+, Max, Business, and Enterprise users across VS Code, Visual Studio, the Copilot CLI and coding agent, github.com, and other supported clients. Its rollout is gradual, so an absent model-picker entry today does not necessarily mean a plan is excluded.
Business and Enterprise administrators should check the model policy: GitHub says new models follow the default enablement setting unless the organization disables that default or this model specifically. GitHub bills usage at provider list pricing under its usage-based billing, which is a separate surface from a direct Claude API account.
What to test before switching
Start with a representative bug fix or small multi-file feature rather than a benchmark headline. Compare the completed change, test results, interventions, total tokens, and wall-clock time against your current model. A cheaper task in Anthropic's tests may not be cheaper in a workflow with different prompts, caching, or tool behavior.
If your Claude Platform integration runs Sonnet with thinking off, Anthropic says it must move to the new between_tools setting for Sonnet 5.5. The launch also includes cyber safeguards: Anthropic says some higher-risk requests can fall back to Sonnet 5, while routine software development is unaffected. That distinction matters when evaluating why two otherwise similar requests behave differently.
Sources
Try the related loot
Give Coding Agents a Free UI Design and PR Finish Gate with UIZZE