pull down to refresh

Surprised no one posted this yet. I haven’t used it but their self-reported benchmarks look good. How many foundation models do we need? Will they begin to specialize and differentiate or will one do it and winner takes all? I hope it’s the former, because it feels very much like Blu-ray vs hd dvd right now.
My working hypothesis is that the AI winner is going to be the one with the best UI/UX plus additional tools and features, and not who has the best model.
reply
50 sats \ 0 replies \ @gmd 19h
Grok actually seems to have a unique moat through its access to twitter data- it's the best place to get immediate updates on breaking news events.
reply
Close to my thesis, but mine is the front-ends like Perplexity / SimTheory / openrouter are going to win.
They are going to win for the following reasons:
  • Model leadership changes often
  • Model will specialize in task or domains (code creation, image creation, music creation, etc)
This is perfect situation for front-ends to flourish. They can intelligently route you to the correct model for the task you are doing
reply
I used it earlier today. Seems the same to me but I didn't ask it anything out of the ordinary. I should ask it to make some images. It wasn't very good at that before.
reply
30 sats \ 0 replies \ @gmd 19h
The last-gen images I've seen floating around twitter are pretty funny- no restrictions creating images of celebrities etc.
reply
Just tell me when you find one that you can't tell the difference between it and non AI content. That's what I'll use.
reply
a good agent wud be one that can tell me which is AI vs non-AI content.
reply