Configure cost and quality in Copilot auto model selection
GitHub Copilot now allows users to select between three distinct tiers—efficiency, balance, and intelligence—for auto model selection. This configuration enables users to prioritize specific trade-offs between computational cost, code quality, and response latency.
Verified State Diff
Impact & Verification Analysis
GitHub Copilot individual and enterprise subscribers.
It provides developers and organizations the ability to balance budget constraints against the need for high-quality code generation, optimizing the developer experience based on specific task complexity.
Full Fact Overview
This update introduces granular control over the underlying LLM routing logic within GitHub Copilot. By exposing these tiers, GitHub is moving away from a 'black box' model selection process, allowing enterprise and individual users to optimize their development environment based on specific project requirements. The 'efficiency' tier likely utilizes smaller, faster models to reduce latency and cost, while the 'intelligence' tier likely leverages larger, more capable models for complex reasoning tasks at the expense of higher latency and resource consumption.