Copilot Auto model selection now available in Visual Studio Code

[nhu-do]

Introducing Copilot Auto model selection 🎊

Not sure which model to use? Auto’s got you covered! 🎉

Today, we’re introducing Auto within VS Code’s GitHub Copilot Chat model picker, now available to all Copilot plans, with rollout to Copilot Business and Enteprise plans starting today, 10/1/2025.

When you select Auto, you’ll be routed to the best available model at any given time. Auto brings you:

• Discounted premium request multipliers for Paid Copilot subscribers
• A mix of premium language models
• Transparency around which model is used and what it costs
• Models that are available based on your subscription and model policies

Auto currently optimizes for model availability and reliability. You’ll still get access to premium models like Claude Sonnet 4, GPT-5, and others that perform well on complex tasks. At any point, you can move off of Auto onto your preferred model.

What’s next? 🚀

We’re just getting started on making your experience better every day. Over time, Auto will factor in your specific task, choosing models that are best for your needs based on a set of criteria. We’ll also expand Auto’s availability to more of your favorite IDEs.

Share your feedback ❤️

Join in the comments below to share your feedback on Auto.

Learn more

• About auto model selection
• VS Code August release notes

[VictorMoni]

Nice 😎👌

[KailashSatkuri-warangal]

Nice ,This is a great addition 👏. Auto model selection will definitely simplify the experience, especially for those of us switching between multiple tasks in VS Code.

Some feedback and suggestions for what could come next:

Task-aware routing: would be amazing if Auto could detect context (e.g., debugging, refactoring, writing docs) and pick the model best suited for that workflow.

Customizable preferences: an option to set priorities (e.g., “favor speed” vs. “favor reasoning depth”) could give users more control without manual switching.

Usage insights: a simple dashboard showing which models Auto routed to, how often, and the impact on performance/cost.

Broader IDE support: JetBrains and Neovim would be high on my wish list.

[nhu-do]

@vlish, excited to share that we just released a more intelligent version of Auto in VS Code today! Now, Auto understands your prompt, using that as an additional signal to choose the most optimal model. Check out our changelog, and our new GitHub Checkout episode that speaks to how this works under the hood!

[vlish]

Thanks @nhu-do for updates. Is this also applicable for Copilot Cloud Coding Agent?

[vlish]

@nhu-do Could you please clarify on question above?

[nhu-do]

Yes, Auto is supported on Copilot cloud agent. However, the version of Auto in CCA is not the newest version. Rather, it optimizes for model reliability using real time health and system insights. Routing based on intent in CCA will come soon.

[vlish]

"Routing based on intent in CCA will come soon." - @nhu-do is there a any preliminary planned release date for this feature, taking into account recent changes for the billing model, it looks like important enhancement.

[Aaisan123]

Thanks for the tip!

…

On Mon, 15 Sept, 2025, 11:29 am Satkuri Kailash, ***@***.***> wrote: Nice ,This is a great addition 👏. Auto model selection will definitely simplify the experience, especially for those of us switching between multiple tasks in VS Code. Some feedback and suggestions for what could come next: Task-aware routing: would be amazing if Auto could detect context (e.g., debugging, refactoring, writing docs) and pick the model best suited for that workflow. Customizable preferences: an option to set priorities (e.g., “favor speed” vs. “favor reasoning depth”) could give users more control without manual switching. Usage insights: a simple dashboard showing which models Auto routed to, how often, and the impact on performance/cost. Broader IDE support: JetBrains and Neovim would be high on my wish list. — Reply to this email directly, view it on GitHub <#173188 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BXKIEMYCZBJ3B64GJBH73ST3SZITZAVCNFSM6AAAAACGIH7OS2VHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTINBQGAYDGMY> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.*** com>

[jennaclarke875]

Nice, thanks for this.

[conradkoh]

May I know if this is available on all accounts or if there is some setting to enable? I have this on my work account but I am not able to access it on my personal one. I assumed it was a rollout thing but it has been more than a week and I still don't see it.

[nhu-do]

Hi @conradkoh, @jennaclarke875 captured most of it, but I will also add that at this time, Auto is only available to individual plans. It will become available to Copilot Business and Enterprise subscribers this week. 🎉

[nhu-do]

👋🏼 Dropping in here to say that I've updated the discussion post as we have now started rolling out to Copilot Business and Enterprise plans 🎉

[conradkoh]

Nice, thanks for the update. I still don't see it on my personal subscription sadly, but looking forward to it!

[nhu-do]

Do you have Telemetry: Telemetry Level set as off on VS Code? If so, that would block Auto from showing up. We're planning to fix this next month :)

To check: Settings > Telemetry > Telemetry: Telemetry Level.

[conradkoh]

Ah, yes I do have that set as off. Thanks for clarifying! :) Will wait for the update next month!

[yeikel]

When can we expect support for JetBrains?

[rajpoot538]

shoaib

…

On Mon, 6 Oct 2025, 5:15 pm Yeikel Santana, ***@***.***> wrote: When can we expect support for JetBrains? — Reply to this email directly, view it on GitHub <#173188 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BSWI4UEPR3D6VDZGJGXDUML3WJMMTAVCNFSM6AAAAACGIH7OS2VHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTINRQGQ2DAOA> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.*** com>