What changed
GitHub Copilot's automatic model selection now offers efficiency, balance and intelligence tiers in VS Code, Copilot CLI and the GitHub Copilot app. Each tier uses the same available model pool but changes how Copilot weighs cost, quality and latency for each prompt; billing follows the model actually selected and paid subscribers retain the 10% auto-selection discount.
Why it matters
Teams can tune an automatic router without pinning a model, but the chosen model, cost and behavior may vary across prompts while plan, administrator, data-residency and FedRAMP policies alter the eligible pool. A tier label is therefore a routing preference rather than a reproducible model identity or budget ceiling.
What to do
Canary all three tiers on the same representative tasks and record the resolved model for every response, accepted output, latency, billed usage and retry behavior. Confirm administrator exclusions and residency constraints, set an external spending limit, and pin a model wherever reproducibility or approval evidence matters.
Caveat
GitHub describes the routing logic and intended tradeoffs, but publishes no independent task-level evaluation for the new tiers. Efficiency does not guarantee a lowest-cost completion, intelligence can still choose a smaller model, and rollout is not yet complete across the supported clients.