GitHub Copilot Fusion Is Now GA: How Its Model Routing and Local-Model Preview Work
GitHub Copilot Fusion is now generally available: instead of picking a model yourself, you choose an effort level (Standard, High or Max) and Copilot Fusion routes and orchestrates models for each task. You're billed per model used at its standard rate, with a limited-time 30% discount on Copilot Fusion model costs for paid Copilot plans. A new experimental preview also lets Copilot Fusion route to on-device local models in Copilot CLI, the Copilot app, and VS Code. [1] [2]
SeekVero editorial · Published · Prepared with AI assistance; no hands-on testing claimed.
Prepared from cited sources. No hands-on testing claimed.
What Is GitHub Copilot Fusion?
GitHub Copilot Fusion is generally available. Instead of picking a model for each task, you select Copilot Fusion in the model picker and let Copilot intelligently route and orchestrate which models it uses. [1]
Copilot Fusion combines Project HydraFusion's multi-model orchestration with Copilot auto model selection's fine-tuned model selection, aiming to balance performance and cost efficiency. [1]
GitHub reports that on the Artificial Analysis Coding Agent Index, Copilot Fusion Max scored 60, within two points of Claude Code running Opus 5.5 Max, while using about half the average API cost per task, about half as many tokens, and about 40% less time per task. [1]
What Are the Copilot Fusion Effort Levels?
When you select Copilot Fusion, you aren't choosing a specific model; you're choosing a system that routes and orchestrates models for each task, and you pick the effort level that fits the task. [1]
| Option or question | What to know |
|---|---|
| Fusion Standard | Prioritizes versatile models for everyday coding. [1] |
| Fusion High | Prioritizes capable models for challenging tasks and deeper reasoning. [1] |
| Fusion Max | Prioritizes the most capable models for the most demanding tasks and complex reasoning. [1] |
How Does Copilot Fusion Orchestrate Models?
For each prompt, Copilot Fusion decides which models to use and how they work together, orchestrating multiple models through one of three workflows. [1]
| Option or question | What to know |
|---|---|
| Single | One model solves the task directly. [1] |
| Cascade | An efficient model drafts a response; a quality check determines whether to accept it or escalate to a stronger model. [1] |
| Critique | One model drafts an initial response, another model reviews it, and the original model revises. [1] |
What Does Copilot Fusion Cost, and Who Controls It?
When you use Copilot Fusion you're billed for each model it uses, at that model's standard rate. For a limited time, users on paid Copilot plans get a 30% discount on model costs when using Copilot Fusion. [1]
Administrators control which models Copilot Fusion can use through enterprise and organization model policies. They can view which models Copilot Fusion picks from, set each model's effective policy state, and disable a model for Copilot Fusion without disabling it for everyone. Administrators can also set Copilot Fusion, at any effort level, as the default model for Copilot users with Enterprise managed settings. [1]
Can Copilot Fusion Route to Local Models?
An experimental preview lets Copilot Fusion choose between supported on-device models and cloud models in GitHub Copilot CLI, the GitHub Copilot app, and VS Code, without you switching models manually for each request. [2]
Local inference does not make a session offline: Copilot Fusion still makes its routing decisions in the cloud, and local models may have different safety protections than Copilot-provided cloud models. [2]
Beyond the Windows-managed download, Copilot CLI's local routing also works with models you configure yourself, such as MAI Code 1.1 Flash or Qwen3.8 Flash Next, through a local OpenAI-compatible server on Windows, macOS, or Linux; you manage that model's download and inference runtime yourself. [2]
- On a supported Windows device, download and configure MAI Code 1.1 Flash through the Windows ML provider. [2]
- In Copilot CLI, enable experimental mode and open /model to follow the setup prompts. [2]
- 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. [2]
- In VS Code, enable the chat.copilot.localModel.enabled setting. [2]
- Once the local model is ready, select Standard in Copilot Fusion to let Copilot choose between local and cloud models, or select the local model directly for your session. [2]
How to Start Using Copilot Fusion
Copilot Fusion is available on all Copilot plans in Visual Studio Code, the GitHub Copilot app, GitHub Copilot CLI, and JetBrains IDEs, with support for additional clients coming soon. If you previously enabled auto, HydraFusion, or HydraFusion Max in the research preview, your selection automatically transitions to the matching Copilot Fusion effort level. [1]
Frequently asked questions
Is Copilot Fusion a new AI model?
No. Copilot Fusion is a routing and orchestration system that chooses and combines existing models for each task; it is not itself a model. [1]
How much does Copilot Fusion cost?
You're billed for each model Copilot Fusion uses, at that model's standard rate. For a limited time, paid Copilot plans get a 30% discount on model costs when using Copilot Fusion. [1]
Which GitHub Copilot clients support Copilot Fusion?
Copilot Fusion is available on all Copilot plans in Visual Studio Code, the GitHub Copilot app, GitHub Copilot CLI, and JetBrains IDEs, with support for additional clients coming soon. [1]
What happened to my HydraFusion research preview settings?
If you had enabled auto, HydraFusion, or HydraFusion Max in the research preview, your selection automatically transitioned to the corresponding Copilot Fusion effort level. [1]
Can Copilot Fusion run completely offline using local models?
No. Even with the local-model preview enabled, Copilot Fusion still makes its routing decisions in the cloud, so a session using local models is not offline. [2]
Next step
GitHub's own changelog post is the official, authoritative source for how Copilot Fusion works, its billing, and its local-model preview.
Read GitHub's Copilot Fusion changelog