By skyforbes This week, I select the Thinking model and ChatGPT switches to Instant anyway, even when I’m explicitly using another model for any task. I first noticed this on the Android app and then tested it on the web, and it’s the same behavior there. Apparently, this is happening with GPT-5.2, because when I choose Thinking on GPT-5.1, it actually uses the model I selected.
Is this a bug, or has OpenAI also started deciding for us between Instant and Thinking? Is anyone else having this issue?