Honestly I’d put very little stock into this, Grok is mostly trained off ChatGPT output - and very specifically trained on benchmark questions to game its own “results” in intelligence tests.

Wait for it to hit lm-arena and for blind user test scores to rank it.

If the above is true, it’ll blitz *every* other model in *every* category. If it doesn’t, and instead lands behind openai/anthropic/google again, then it’s just another gamed benchmark.

e/acc || Building @ /imgn || Dogecoin core dev 2013/14

product & marketing @teller 





























































hosting /kieran /usdc /loans /disclosure 🛸

The road to hell is paved with good intentions.

💀 

But nah, honestly I don’t mind Elon. Think he’s an example of someone who does genuinely want to make things better for humanity, except uh, sometimes he gets led into doing the opposite (with good intentions) 

I think Grok is v much like the emperor’s new clothes to him. He’s blinded by benchmarks, and has no idea how easily they’re gamed.