Last Updated on October 8, 2026 by Spencer Lee
GitHub Copilot is getting a new ability to choose between local and cloud AI depending on the task. Microsoft says the feature will allow Copilot to use on-device models when they are suitable and cloud-scale models when more capability is needed. The move could make laptop hardware increasingly important for AI-assisted software development.
What is GitHub Copilot changing?
Microsoft says GitHub Copilot will be able to determine whether a coding task is better handled by on-device intelligence or cloud-scale models.
The goal is to give developers more control over the trade-off between performance, latency and cloud computing costs. Microsoft says the capability is coming soon.
This is different from simply giving developers the option to select a local model manually. Copilot is being designed to help determine which type of model is appropriate for a particular task.
Can GitHub Copilot already use local models?
Yes.
GitHub Copilot CLI can now discover supported local models from a running Ollama installation. Developers can use the /model command to find available local models and select one for their current session.
GitHub also says the local model must already be installed, and the model needs to support tool calling and streaming.
The new discovery feature does not automatically download or install models.
Why does this matter for laptop hardware?
Local AI shifts some of the computing workload from remote servers to the user’s own machine.
That makes hardware such as RAM, GPU memory, CPU performance and AI acceleration more relevant to developers who want to run models locally.
Microsoft’s Windows ML platform is designed to run local AI workloads across supported CPUs, GPUs and NPUs from AMD, Intel, NVIDIA and Qualcomm. Microsoft has also added experimental support for GGUF models through llama.cpp in Windows ML.
Does local Copilot mean cloud AI is going away?
No.
Microsoft’s announcement specifically describes a hybrid approach. Copilot will be able to use local intelligence when appropriate while continuing to use cloud-scale models for tasks that require more capability.
That means developers don’t have to choose between local and cloud AI for every task.
Instead, Microsoft’s stated goal is to route work according to the requirements of the task.
What does this mean for AI laptops?
The change could make local AI capability a more practical consideration when buying a laptop for software development.
A developer who only uses cloud-based Copilot does not necessarily need a powerful AI laptop.
But developers who increasingly use local models may benefit from more system memory, stronger GPUs or supported local AI acceleration.
The actual hardware requirements will depend on the models and workloads being run, so Copilot’s new routing feature should not be interpreted as meaning that every developer needs a high-end AI laptop.
Is local AI coding available on Windows now?
Parts of the local AI development stack are already available.
Microsoft’s Windows ML provides a framework for local inference, while GitHub Copilot supports local model discovery through Ollama in its CLI. GitHub also made local sandboxing generally available for Copilot CLI, the Copilot app and supported VS Code sessions using Agent Host.
The automatic decision-making between local and cloud models is the part Microsoft says is coming.
What happens next?
Microsoft says GitHub Copilot’s intelligent local/cloud model selection is coming soon.
The important development for laptop users will be seeing which models Copilot can run locally, what hardware those models require, and how much of a developer’s workload can realistically be handled without cloud inference.
Also read: Mac vs. Windows for AI Work in 2026: Which Is Better?
As local AI becomes more important to software development, the operating system and hardware platform can affect which models, frameworks and acceleration technologies developers can use.
