GitHub Copilot Chat: Auto Model Selection Now Available to All Users
Auto model selection is now generally available in Copilot Chat on github.com and the GitHub mobile app for all Copilot plans, extending the rollout that previously reached VS Code, JetBrains, and the Copilot CLI. Auto routes each request to the optimal model based on task complexity and real-time model health, choosing across Claude Sonnet 4.6, GPT-5.4 mini, GPT-5.4, and Haiku 4.5 depending on your plan and policies. Paid subscribers get a 10% billing discount when using auto, can hover a response to see which model answered, and can still switch to any specific model per response.
Sources & Mentions
5 external resources covering this update
Auto mode in Copilot Chat available for all users
GitHub Changelog
About Copilot auto model selection
GitHub Docs
Auto model selection is generally available in GitHub Copilot in Visual Studio Code
GitHub Changelog
GitHub Copilot CLI now supports Copilot auto model selection
GitHub Changelog
Supported AI models in GitHub Copilot
GitHub Docs
Auto mode comes to github.com and mobile
GitHub has made auto model selection generally available in Copilot Chat on github.com and in the GitHub mobile app, for all Copilot plans. This extends the auto mode rollout that previously reached Visual Studio Code, JetBrains IDEs, and the Copilot CLI, bringing intelligent model routing to two of the most widely used chat surfaces.
The idea behind auto is simple: rather than choosing a model yourself for every question, you let Copilot pick the best one for the job.
How auto routing works
Auto mode chooses the optimal model based on two signals working together. One system tracks real-time model health and availability, while the other evaluates the complexity of your task. Combining those, Copilot routes each request to the model best suited to handle it, balancing capability against current reliability.
On github.com and mobile, auto can select across Claude Sonnet 4.6, GPT-5.4 mini, GPT-5.4, and Haiku 4.5, based on your plan and policies. Lighter requests can be served by faster, lower-cost models, while harder tasks can be routed to more capable ones, without you having to make that call each time.
Transparency and control
See which model answered
Auto does not hide its choices. You can see which model was used for any response by hovering over that response, so the routing stays inspectable rather than opaque.
Manual switching still available
Auto is the default convenience, not a lock-in. You can switch between auto and any specific model of your choice on a per-response basis, so if you want a particular model for a given question, you can still pick it.
Billing benefit
Using auto comes with a pricing incentive: for all paid subscribers, requests served through auto mode are billed at a 10% discount. In practice the discount applies to the model multiplier, so a request that would normally draw down one premium request instead draws down slightly less. That makes auto both the easy choice and the cheaper one for everyday use.
Why it matters
Model choice has become a real cognitive tax: developers juggle several frontier models with different strengths, costs, and availability. Auto mode offloads that decision to a routing system that factors in both task difficulty and live model health, then sweetens it with a discount and keeps the choice transparent and overridable. Bringing it to github.com and mobile means the convenience now follows you across the surfaces where you actually work, not just inside the IDE.