GitHub Copilot Fusion Goes GA: Intelligent Model Routing and Orchestration Now Available

GitHub has announced that GitHub Copilot Fusion is now generally available, introducing a new paradigm where developers no longer need to manually select individual AI models for specific coding tasks. Instead, users select Copilot Fusion in the model picker and let the system intelligently route and orchestrate the appropriate models based on the task's requirements.

Overview of GitHub Copilot Fusion

The new feature integrates Project HydraFusion’s multi-model orchestration capabilities with Copilot’s auto model selection, combining their strengths to balance performance and cost efficiency. Rather than choosing a single model for a task, Copilot Fusion dynamically determines which models to use and how they collaborate, allowing developers to simply choose the right level of effort.

Performance and Cost Efficiency

According to GitHub, Copilot Fusion Max achieved a score of 60 on the Artificial Analysis Coding Agent Index, positioning itself within two points of Claude Code with Opus 5.5 Max. Notably, this performance was delivered at approximately half the average API cost per task, using roughly half as many tokens, and completing tasks in about 40% less time. These metrics highlight the system's ability to optimize both speed and expenditure.

How Copilot Fusion Works

When a user selects Copilot Fusion, they choose an effort level—Standard, High, or Max—without specifying a particular model. The system then analyzes each prompt and selects the optimal models and workflows. Three primary orchestration workflows are utilized when needed:

• Single: One model solves the task directly.
• Cascade: An efficient model drafts a response, followed by a quality check that either accepts the draft or escalates it to a stronger model.
• Critique: One model drafts an initial response, another reviews it, and the original model revises it.

Administrative Controls

Enterprise administrators retain significant control over Copilot Fusion. They can dictate which models the system is permitted to use based on their organization's model policies. Administrators can independently set the effective policy state for each model and disable models specifically for Copilot Fusion without affecting the broader organization. Furthermore, Enterprise managed settings allow administrators to set Copilot Fusion as the default model at any chosen effort level.

Pricing and Availability

Users are billed individually for each model used at its standard rate. For a limited time, users on paid Copilot plans will receive a 30% discount on these model costs. Currently, Copilot Fusion is available across all Copilot plans in Visual Studio Code, the GitHub Copilot app, GitHub Copilot CLI, and JetBrains IDEs. Support for additional clients is scheduled for future release.

Industry Reaction

Hari Pokhrel, Principal Software Engineer at AT&T, shared his experience with the tool: "GitHub Copilot Fusion Standard has improved token efficiency across AT&T by automatically selecting the right model for the task, reducing unnecessary spend and preserving capacity for higher-value work. Previously, developers would often choose a more powerful model for a complex task and forget to switch back for everyday activities. Copilot Fusion eliminates that friction, optimizing model selection without requiring developers to manage it themselves."

How to Verify

To test the feature, developers can update their client software, open the model picker, select Copilot Fusion, and choose an effort level. Those who previously enabled auto, HydraFusion, or HydraFusion Max configurations will find their settings automatically transition to the corresponding Copilot Fusion effort levels.