The problem is when they stop releasing big models for so long that their small models overtake large ones.
replyUser wise, the average person will see pro and expect it to be better. Where as companies like OpenAI have tended to release across small/mid/large such that the disparity hasn’t been large in terms of a smaller model outperforming their latest mid/large model.
You would hope they wouldn't have old big models that are worse than new small models available in the interface like that.
replyPro is better, except for agentic coding work
reply