GitHub Copilot CLI makes local Ollama models easier to choose

GitHub Copilot CLI now lets you find compatible local Ollama models in its model picker and use one without restarting your session.

Choose a model without restarting

Starting with version 1.0.94-0, type /model inside GitHub Copilot CLI to see supported models from an already-running local Ollama instance alongside your configured models and GitHub Copilot’s cloud models. Local and external models were already supported through BYOK (Bring Your Own Key) provider configuration; the new feature is discovery in the picker.

Discovery does not add models automatically: select one, review its provider and endpoint, and confirm “Add and use for this session” or “Add without switching.” You can use the selected model in the current session without restarting the CLI.

Have Ollama and a compatible model ready

Ollama and the model you want must already be installed. The discovery flow does not install the runtime or download model weights.

Models need both tool calling, also known as function calling, and streaming support. If a provider connection fails, the picker displays an explanation to help you identify what needs attention.

Local inference is not an offline switch

Choosing a local model does not enable offline mode or disable GitHub telemetry. To prevent the CLI from contacting GitHub’s servers, set COPILOT_OFFLINE=true separately; full network isolation also requires a provider that is local or within the same isolated environment. If you configure a remote provider, your prompts and code context still go to that provider over the network, even with the offline flag set.

Microsoft has also announced intelligent Auto orchestration that can route work between local and cloud models, with availability planned by the end of October 2026. That is a separate upcoming feature, not something /model enables today.