It seems to write ‘War and Peace’ for the simplest of refactorings and thus is much slower than Claude, is there a way to turn off the thinking and just get the answer?
deepseeks USP is that thinking lol i dont think its possible
you can use gemini-pro or o3-mini instead they’re moderately good at reasoning while being faster than r1 or v3
Hey, if we hide it, you’ll still wait the same amount of time while the reasoning is happening. It’s not possible to disable this at the model level since it’s a reasoning model, and that’s how it works. We might hide it in the future but will allow you to reveal it if you need to.
a toggle to show or hide the think
tag content would be ideal - or output it to its own pill that is collapsed by default that can be expanded in desired.
I definitely read the thinking as it shows where it goes wrong when it does so.
I also find value in reading it, as it sometimes reveals insights I hadn’t considered.