GitHub has introduced a new 'Auto' model selection feature for Copilot users in Visual Studio Code. This update is available for developers using Copilot Free, Pro, and Pro+ subscriptions, with plans to extend access to Copilot Enterprise and Business subscribers soon. The Auto model selector is integrated within the Copilot Chat interface, allowing users to let GitHub Copilot choose from several large language models, including GPT-5, GPT-5 mini, GPT-4.1, Sonnet 4, and Sonnet 3.5. Currently in preview, this feature aims to route requests to the most available model, with future plans to make selections more context-aware based on user tasks.
Developers can monitor which model was used for any response by hovering over the output, and they retain the ability to switch back to manual model selection at any time. All user settings and policies are respected during Auto selection. For those on Pro and Pro+ plans, there's a 10% reduction in premium request usage when Auto picks a model with a 1x multiplier, reducing costs for frequent users.
GitHub, the company behind Copilot, continues to expand its AI-assisted coding tools, aiming to streamline coding processes and reduce friction in model selection. By introducing Auto, GitHub responds to users seeking efficiency and flexibility, especially as more language models become available. Early feedback from developers highlights the convenience of not having to manually choose a model while maintaining transparency and control.
