Some good, if brutal, points here. There's no question xAI's not in the same league with the big ones (does anyone not on Twitter ever even think about it?), but losing ten of your twelve founders in two years just doesn't sound good from any angle.
The timing is remarkable. Tesla disclosed the $2 billion xAI investment in its Q4 2025 earnings report on January 28. Days later, SpaceX acquired xAI in a deal valuing the combined entity at $1.25 trillion. Now Musk is telling the world the thing he just sold to his own public and private investors was broken.
I do. It's one of my fallback models. Next to undressing people, it's still good at search (for the insane things I prompt) and prompt-writing.
Edit: just checked. It's ranked 3rd on Arena right now for generic chat:
and 2nd for Search:
Interesting, although I notice that's the beta (and tagged with a "preliminary" mark). I'm not on Twitter, and it's just not something I ever see anywhere in the AI discussions or searches I do.
I never touch AI for search, other than when I use NotebookLM's built-in option (though obviously, there's search that happens implicitly based on some prompts), so I admit that's just not a need I think about much.
Is Grok good at prompt-writing outside of Grok itself? I've generally stuck to the model I'm using to get prompt tips (so Gemini for Gemini, etc).
imho, yes. For example, I feed it
claude-opus-4.6-thinkingthought process slop to identify where it went wrong, and to give me an integrated one-shot prompt. This often works. The only thing it has trouble with is Qwen3.5 but especially on the smaller models, the reasoning crap the latter puts out is ridiculous.I always do the opposite. Second opinion > repeated opinion.
At least admitting it is a step. A lot of companies would just keep pushing forward and pretend everything is fine.