Allie K. Miller
I tested Astra early. What I would want EVERY business professional to know about the brand-new model from OpenAI: 1) Computer use and browser use are just incredible. I'm confident that pretty much any stable workflow done on a computer can be at least partially done by AI (and yes, I realize that description has a lot of hedging but I'm sure there are edge cases I didn't explore). It's just so so good. Imagine you were a creepy boss and recorded your screen for 7 straight days...what would you have AI do that you've been doing? Can you challenge yourself to go mouse-free for a day? 2) Still failed my complex word play challenges. No model has passed it yet. 3) Didn't piss me off. I cannot overemphasize this enough. I have been screaming at Claude more and more these last few weeks because of its convoluted extremely technical responses sometimes (so much so, that I adapted a new /wtf skill to deal with it AND as a ratio of my overall AI work, Codex has shot up - I'm using Claude Code desktop app less and less). I used to say Claude was a better writer and I no longer say that. 4) Codex is a better daily driver. 5) I'm not worried about tokens like I am with the Fable family (important to note that I'm also not really worried about tokens with Opus). 6) Others tested a lot around 3D and animations. I didn't, but they seem v happy with the results. I watched their results and wasn't as stunned, but that is OK. I don't want to see worlds from scratch, I want to see accurate worlds from reference points. I know AI can generate from scratch, but that's not what most work starts as. 7) Would probably still go to Fable 5.1 for "mega architect build" type asks. And unrelated but thank you to the OpenAI engineer that added my request of Codex voice telling you when you had to approve a request. We appreciate when the labs listen 🤘 I shall continue to hammer the hell out of it and provide business takeaways. Engineers, go follow @clairevo and @danshipper for their additional takes.