Configure cost and quality in Copilot auto model selection
TL;DR
GitHub Copilot's auto model selection now offers three tiers: efficiency, balance and intelligence. The chosen tier sets how auto weighs cost, response quality and speed when it picks a model. That matters most for teams that already use auto broadly but want tighter control over spend or latency. GitHub has not yet detailed which models sit behind each tier or how large the differences are in daily work.
Nauti's Take
Small teams should first run the same real-world tasks across all three tiers and measure quality, latency, and usage. Before setting a default, verify whether GitHub clearly documents the model mapping and resulting costs for each tier, since the practical difference between the options remains untested from the announcement alone.
Sources