重度用户对比 ChatGPT Dot、Claude Code 与 Grok Bot

David Shapiro (L/0) @DaveShapi原推发布于 2026.10.18

在 X 上打开原推

I've basically stopped using ChatGPT Dot for most things. It's useful for bursting several parallel tasks, for instance, researching and designing YouTube thumbnails. When the deliverable is that discrete and concrete, Dot does pretty well. It's also nice to have a single touchpoint for other threads, emails, business files, and that sort of thing. But it regularly fails to provide updates, and doesn't report when things fail. So, on ChatGPT, I've reverted mostly to using ordinary Work sessions and Pro sessions, which is mostly what I do with Claude Code as well - dedicated threads for specific work and business functions. Grok Bot would be a top pick since it is explicit about which bots are running, but the biggest problem is that it obscures which model you're using behind a completely opaque router and refuses to tell you OR give you any control over it. So Grok Bot is frequently (still) too idiotic to use. With that said, Grok Bot was good for brainstorming concept art for a cyberpunk game I've got in the back of my mind, but so far, that is the most use I've gotten out of it. I've got all these models connected to my email, local file system, calendar, and Google Drive, so there's no feature difference there. Well, I removed Grok Bot from all integrations because it was just too stupid, and kept asking permission even after I explicitly whitelisted stuff three times. I figured if it was too stupid to keep track of that permission, I had zero trust it wouldn't fuck up everything else. Someone said last year "the only people impressed by Grok seem to be those who haven't used any other tools" and I wonder if that's still true. For those who've gone all in on Claude Code, ChatGPT Desktop/Dot, and Grok Bot, but still prefer Grok Bot... why?