GitHub Copilot CLI can now discover your local Ollama models
Version 1.0.94-0 adds a model picker that finds running Ollama instances alongside cloud models, though it stops short of offline mode.
GitHub Copilot CLI 1.0.94-0, shipped 7 October, can now discover models from a running Ollama instance and list them alongside your configured and cloud-based models. Run /model and the picker shows what’s available, with an option to “Add and use for this session” or “Add without switching” — no restart required.
The feature only discovers models you’ve already set up: Ollama and the models themselves still need installing yourself, this doesn’t pull anything automatically. Models must support tool calling and streaming to work in the CLI. If a provider connection fails, the picker shows why inline rather than silently omitting it.
Why it matters: Copilot CLI already supported bring-your-own-key and local providers through environment variables, but switching meant editing config and restarting. A picker that finds what’s already running locally removes that friction for anyone testing a local model against Copilot’s cloud defaults mid-session.
The caveat: selecting a local model doesn’t enable offline mode or stop GitHub telemetry. That still needs COPILOT_OFFLINE=true set explicitly — this change is about convenience, not air-gapping.