Gemini's Auto Mode Is the Right Default and the Wrong Habit

Google's Gemini app now offers an Auto setting that picks the model per prompt, plus three thinking tiers. On the free tier it auto-selects and quietly drops to a lighter model for most queries.

For casual use that's perfect -- you stop paying attention to which brain is answering and just get a result. Most questions don't need the biggest model.

The risk is habit. When I do need depth -- a long contract, a real coding task -- Auto's quiet downgrade bites me, and I forget to force the strong model.

My rule: Auto for the daily stuff, manual for anything where the answer has consequences. The convenience is real; the amnesia it causes is the cost.