current round of frontier llms has some of the best personalities since og gpt-3
asked to solve a problem, gpt-5 instead wrote out a proof that it's impossible, while sonnet went ahead and solved it while only barely stopping short of telling me my different proposed solution was dumb