Gotta love techbros using Grok's eval scores (which is now called out as manipulated) to try to meme Grok into existence 😂

The most expensively trained model by far, and still worse than Sonnet in coding, still worse than DeepSeek in intelligence, still lower usage than all other relevant models.

Why does everything Elon touches fall apart lmao

https://chainleft.art

Dad, data scientist, blockchain native artist.

Power consolidation=evil.

Grok hasn’t yet developed consistent rules that produce satisfying outputs, but it’s entirely unique in its own way.

What's unique about it? Just lower guardrails with extra censorship, no?

Other AIs respond based on the data and public information they’ve been trained on unless we request a Google search. Grok is continuously integrated with various tweets on Twitter, giving it a broader range of responses. Its answers also tend to be more humorous and less formal. Because of this, it requires more powerful processing and additional rules.
With these conditions, it’s not even comparable to ChatGPT.

so, just including recent casts in the prompt .. we got a similar behaviour.. it does not look too complex nor special.. to justify the costs.. 

that said, have no tried it