I've been experimenting with ChatGPT and Claude Sonnet 4's deep thinking modes. For sure they can pump out content, 20 page documents without breaking a sweat. But is that really "Deep" or just spending more time and energy? Frankly, it is hard to get a 20 page doc and then have to spend time analyzing whether it is good. I ran two, one in ChatGPT 4o and one with Claude and have 38 pages of.... "Stuff". Hmmm.