GitHub Copilot has expanded its GPT‑6 lineup with two new models—Sol and Luna—joining the previously available Astra. The change gives engineers a direct way to trade off model capability, response speed, and cost, while also introducing new policy controls and usage‑based billing that affect platform and security operations.
Model differences and selection criteria
- GPT‑6 Sol: Positioned as a balanced, all‑round model suitable for interactive coding sessions and agentic workflows that require careful, multi‑step validation.
- GPT‑6 Luna: Described as a lightweight, low‑cost option aimed at smaller, faster tasks, making it the cheapest model in the GPT‑6 family.
- Existing Astra: Remains available for scenarios that need the original GPT‑6 capabilities.
Practitioners can now pick the model that best matches a given workload, whether that is a complex refactoring operation (Sol) or a quick autocomplete request (Luna). The model picker is exposed in all major Copilot entry points, including VS Code, Visual Studio, the Copilot CLI, the cloud agent, the desktop and mobile apps, JetBrains IDEs, Xcode, and Eclipse.
Impact on deployment and operations
All three models are billed on a usage‑based basis, so cost monitoring must be incorporated into any CI/CD pipeline or automated workflow that invokes Copilot. The rollout is gradual, meaning some environments may not see the new models immediately; teams should verify availability in their chosen IDEs or agents before updating automation scripts.
Access to the models depends on the Copilot subscription tier. Sol is limited to Pro+, Max, Business, and Enterprise plans, while Luna is also available to the base Pro plan. This tiering influences budgeting and licensing decisions for platform teams.
Security and governance considerations
Enterprise and Business administrators can manage model exposure through a model policy in Copilot settings. By default, new models are enabled automatically unless an admin disables the global default or explicitly blocks a model. Organizations should review these policies to ensure that only approved models are used in production pipelines, especially where cost or data‑handling requirements differ between Sol and Luna.
Because usage is tracked per request, audit logs can be leveraged to correlate model selection with downstream code changes, supporting compliance and incident‑response workflows.
Related CloudNinjas coverage: AI engineering.
What This Means For Practitioners
Evaluate which workloads benefit from Sol’s validation focus versus Luna’s speed and cost profile, and adjust CI/CD scripts to select the appropriate model via the picker API. Update subscription and policy settings to align model access with team budgets and security guidelines. Finally, monitor the gradual rollout and usage metrics to confirm that the new models are available where needed and that billing aligns with expectations.
