GPT-5.2 is live

[zemaj]

Seeing great results so far. Tool use with this model is A.M.A.Z.I.N.G. this is particularly important for Code since we push the range of tools available a bit further than most CLIs.

I recommend using GPT-5.2 High for most tasks - it's a great balance of speed and intelligence - it's even more token efficient than -max!

[cbusillo]

Do you use any local models to save on rate-limit or API costs? I have a Mac with 128GB of RAM. Is there a workflow with Code that I could use? Or do you think its not worth it?