What is the common knowledge on the best models to use? I keep returning to Sonnet, but Gpt-4o seems to be giving better results on some files.
Experiment:
Launch multiple Cusro sessions - set the ONLY model in each session to only one model, sonnet, gpt, 01, gemini,
Feed them all the same prompt, a medium level complexity multi step and phase prompt problem and see who can perform best WRT how neatly packaged the entire experience is and the resuktant output. YOLO all of them.