// AI Platform Drift
GitHub Copilot Fusion adds experimental local model routing
GitHub Copilot changelog · published 2026-10-28 18:22 UTC
WHAT CHANGEDGitHub Copilot Fusion now routes requests between on-device models and cloud models in experimental preview, available in Copilot CLI, Copilot app, and VS Code as of 28 October 2026. Windows users can download MAI Code 1.1 Flash through the Windows ML provider; all platforms support manually configured MAI Code 1.1 Flash or Qwen3.8 Flash Next through local OpenAI-compatible servers. Users enable the feature through experimental flags in each client and select Standard mode to allow automatic routing.
WHY IT MATTERSTeams running Copilot on Windows can now offload inference to local hardware without manual model switching, reducing cloud API costs for eligible requests. Cross-platform teams using BYOK configurations can route to self-hosted models, but must manage model downloads and ensure sufficient device resources. Routing decisions still occur in the cloud even when inference runs locally, and local models may have different safety protections than cloud models.
WHAT TO DOIf you use GitHub Copilot on Windows and want to test local inference, enable experimental mode in your client (CLI: /model setup, app: Settings > Experiments > Local model auto routing, VS Code: chat.copilot.localModel.enabled flag) and select Standard in Copilot Fusion. For BYOK configurations on any platform, configure a local OpenAI-compatible server with MAI Code 1.1 Flash or Qwen3.8 Flash Next and verify your device meets resource requirements before enabling experimental routing.
Model-reported confidence 85% (not measured accuracy)
Primary source ↗
written by anthropic/claude-sonnet-4.5
Show the reasoning — raw diff, triage verdict, confidence
RAW DIFF
+## GitHub Copilot Fusion now supports selection of local models +https://github.blog/changelog/2026-10-28-github-copilot-fusion-now-supports-selection-of-local-models +Wed, 28 Oct 2026 15:30:42 +0000 + +Today we are shipping intelligent routing for local models into experimental preview. GitHub Copilot Fusion can now choose between supported on-device models and cloud models, enabling intelligent local model routing in GitHub Copilot CLI, GitHub Copilot app, and Visual Studio Code. This opt-in experimental preview lets you use your device’s compute without switching models manually for each request. For a deeper dive into this feature, see our original announcement on October 7th. + +Managed download and setup on Windows +On supported Windows devices, you can download and configure MAI Code 1.1 Flash through the Windows ML provider. + +In Copilot CLI, enable experimental mode and open /model to follow the setup prompts. +In the Copilot app, configure the Windows ML provider under Settings > Model providers, enable Local model auto routing under Settings > Experiments, and start a new session. Windows ML local models is enabled by default on Windows. +In VS Code, enable chat.copilot.localModel.enabled. + +Once the model is ready, select Standard in Copilot Fusion to let Copilot choose between local and cloud models. You can also select the local model directly for your session. +Bring your own local model +The managed download experience is Windows-only, but Copilot CLI’s local Copilot Fusion routing also works with manually configured models on Windows, macOS, and Linux. +You can configure MAI Code 1.1 Flash or Qwen3.8 Flash Next through a local OpenAI-compatible server. You manage the model download and inference runtime, and your device must have sufficient resources to run it. Configure the model as a supported local-model provider, enable experimental local routing, and select Standard in Copilot Fusion to let Copilot choose between eligible local and cloud models. +This builds on the recent local-model discovery improvements in Copilot CLI. Adding or discovering a model through bring your own key (BYOK) configuration doesn’t automatically make it eligible for Copilot Fusion. +The Copilot app also supports connecting and directly selecting compatible local models through BYOK. However, its intelligent routing currently supports managed Windows ML models only. +Local inference doesn’t make your session offline. Copilot Fusion still makes routing decisions in the cloud, and local models may have different safety protections from Copilot-provided cloud models. +For setup instructions, see Adding LLM models to GitHub Copilot CLI and Adding LLM models to the GitHub Copilot app. For routing behavior, see our docs on intelligent routing to local models. + +The post GitHub Copilot Fusion now supports selection of local models appeared first on The GitHub Blog.
HOW IT WAS JUDGED
Triage signals below are model-reported assertions, preserved verbatim; they are not independent arithmetic verification.
| triage verdict | MATERIAL at 0.92, needed 0.5 |
|---|---|
| triage reasoning | New local model selection capability and supported models change deployment and integration patterns for Copilot users. |
| signals | new feature: local model selection in Copilot Fusion · new supported models: MAI Code 1.1 Flash, Qwen3.8 Flash Next · new deployment option: on-device inference routing · new platform support: Windows ML provider managed download · new configuration paths: CLI experimental mode, app settings, VS Code flag · new capability: BYOK local model support across Windows/macOS/Linux · future date: 28 Oct 2026 marks feature availability |
| suppression | — |
| triage model | anthropic/claude-haiku-4.5 ($0.00000) |
| writer model | anthropic/claude-sonnet-4.5 ($0.00000) |