
On May 13, a Copilot cloud agent release introduced an "Auto" option in the model picker that routes requests to the best available model based on system health and model performance. When users select Auto, the cloud agent applies a 10% discount on the normal model multiplier and exempts those requests from weekly rate limits. The same release also added the ability to start Copilot cloud agent tasks via the REST API.
The Auto rollout is part of several May changelog entries that adjusted model support and telemetry. Recent notes include retirements or deprecations of Grok Code Fast 1, GPT‑4.1 and Claude Sonnet 4, along with functional improvements such as Copilot code review comment updates, expanded Rubber Duck support across more models, and new usage metrics for code review comment types.
For development teams, Auto selection should simplify model routing and help lower costs through the stated 10% multiplier discount while avoiding weekly rate‑limit interruptions. Administrators should review the auto model selection documentation plus recent usage, metrics and billing updates to prepare for any operational or cost impacts.
Sources
Replies (0)
No replies in this topic yet.