Source item
Open original source ↗After extensive evaluations (vibes) O1 Pro is much better than O1 Preview.
The OpenAI people saying the evals are an early checkpoint are not lying.
Quoted context
These are the quoted author’s words.
O1 released today is worse than O1 preview in many ways as per OpenAI's own benchmarks What did OpenAI do to mess this up? I'm guessing over quantize or decrease the thinking time generally to make it more servable, but not sure... https://t.co/OUV7SxIkTF
dylan522p ↗