GitHub announced general availability of GPT-6.1 Terra in Copilot on October 26. Its positioning is straightforward: everyday terminal and coding work at lower cost than the larger Sol model. GitHub says it performed better than GPT-5.6 Terra on some early terminal and C# coding evaluations, though those are vendor tests rather than a result on your own repositories.
The model appears in supported Copilot interfaces including VS Code, CLI, GitHub.com and Xcode, with gradual rollout. Usage-based billing applies at provider list prices, and organizational model policies may affect availability.
Your best benchmark is a small, verifiable task: modify one test, update a dependency or fix a clearly isolated bug. Compare total time, number of corrections and billed usage against the model you already use. The cheapest per-token model is not automatically cheapest per completed feature.
Run the same well-scoped coding task with Terra and your usual model, and record quality, elapsed time and cost.