GitHub Copilot has introduced new configuration options for its auto model selection feature, effective September 14, 2026. Users can now choose from three distinct tiers: efficiency, balance, and intelligence. These tiers enable customization of how Copilot's auto-selection process prioritizes cost, quality, and response time when responding to prompts.
Key Points
- GitHub Copilot auto model selection now offers three tiers: efficiency, balance, and intelligence.
- Users select a tier to reflect how auto model selection weighs cost, quality, and response time for each prompt.
- Copilot optimizes model selection based on the chosen tier.
- All three tiers utilize the same set of available models.
- Auto model selection individually evaluates each prompt to select a best-suited model.
- Usage is charged based on the model auto selects, regardless of the chosen tier.
- Paid subscribers receive a 10% discount on usage billed through auto.
- The feature is rolling out in Visual Studio Code, Copilot CLI, and the GitHub Copilot app.
Context
According to GitHub, the efficiency tier prioritizes low costs and is suitable for fast, straightforward tasks. The balance tier weighs cost, quality, and latency together, making it appropriate for everyday work. The intelligence tier prioritizes quality and is designed for complex tasks. This update represents an initial step toward greater customization and transparency in model selection tradeoffs, according to GitHub.
Why It Matters
This update provides developers with more control over the performance characteristics of GitHub Copilot, allowing them to align AI assistance with specific task requirements and budget considerations. The ability to choose between efficiency, balance, and intelligence tiers can help optimize resource use and output quality for different coding scenarios.
What To Do
- Check for the new auto model selection tiers in Visual Studio Code, Copilot CLI, or the GitHub Copilot app.
- Test the efficiency tier for simple tasks like adding docstrings to evaluate cost savings.
- Experiment with the intelligence tier for complex coding challenges to assess quality improvements.
- Compare usage charges across tiers to understand the financial implications of each setting.