> Deep Think

Deep Think is no longer a mode. Every model thinks in agentic steps, and thinking effort gives you finer control than the mode ever did.

What it did

It was a mode you turned on to get a harder, more careful answer than the default would give.

What to use now

That is now a matter of raising thinking effort, which applies per step on any model. More control over the same thing, rather than a separate switch.

See Agentic steps for the full picture.