How to disable (hide) thinking output from model response?

[vec715]

how do i hide the thinking output from the model response? i connected the nvidia/nemotron-3-nano model and the copilot chat plugin keeps showing the thinking process, which is driving me crazy because i have to manually figure out where the actual answer starts and where it's still thinking. anyone know how to turn that off?

[logancyang]

Looks like this model doesn't output thinking as wrapped in <think> or in a recognized field, hence Copilot has no way to recognize its thinking tokens. You can either use a model that respects the thinking format of openai, claude or openrouter, or make it respect it if you are up for a technical challenge. Copilot isn't the layer to blame here.