GitHub has launched an Auto model selection option for Copilot in JetBrains IDEs, Xcode, and Eclipse, now available in public preview. The feature removes the guesswork from model selection by automatically routing requests to the most appropriate available model—and it comes with a cost benefit for premium request users.
Let Copilot Choose for You#
The Auto option appears in the Copilot Chat model picker alongside individual model selections. When enabled, Copilot intelligently routes your requests based on model availability, currently selecting from GPT-5, GPT-5 mini, GPT-4.1, Sonnet 4.5, and Haiku 4.5 depending on your subscription type.
Transparency remains built into the experience. Users can hover over any model response to see exactly which model handled their request, eliminating any black-box concerns. The feature also respects all existing user model settings and organizational policies.
Premium Request Savings#
For Copilot Pro, Pro+, Business, and Enterprise subscribers, Auto offers a tangible financial benefit: a 10% discount on premium request multipliers. When Auto selects a model with a 1x multiplier, for example, you’ll only consume 0.9 premium requests instead of the full amount.
The current preview focuses on models with 0x to 1x multipliers, keeping costs predictable while GitHub expands the feature’s capabilities.
Getting Started#
Individual users can start immediately by opening Copilot Chat in their IDE and selecting Auto from the model picker. The option is available across all supported plan types.
Copilot Business and Enterprise subscribers need an additional step: an administrator must first enable the Editor preview features policy before team members can access Auto mode.
The Road Ahead#
GitHub is positioning this initial release as a foundation for smarter automation. While the current preview optimizes primarily for model availability, the company is actively developing future updates that will factor in task type and context when selecting models.
The feature represents GitHub’s continued push toward making Copilot more intelligent and autonomous—reducing decision fatigue for developers while potentially lowering costs through optimized model routing.