Solving complex problems with OpenAI o1 models

Any new insights here? Watching now, will update if there's anything of note.