Title says it all. Just got email that we can stream o1 now, but haven’t we all moved onto o3? 
can’t wait till everything gets unified, but until then—what’s an app to do?!
1 Like
The AI world is moving fast. Even faster than the o1 devs.
Maybe they should use o3 
Hah! Getting the email was equivalent to, “hey, we’ve upgraded gpt-3.5-turbo”, because then I have to think about why—in my mind—I’ve essentially shut out o1…and then I go searching and find this:
And then I remember—OH yeah, that’s why I stopped thinking about o1, because with o3-mini medium I can get basically the same thing—for 15x less the cost…
…unless I’m missing something.
1 Like
Vision!
BUT! not everyone needs that 
1 Like
Yeah, especially with the nomenclature I know it’s a bit confusing. o1 is 1) our current full-size reasoning model 2) the only reasoning model that supports vision in the API, so there’s still active development around its capabilities, since we know many developers still rely on it. While o3 is coming soon, the email was sent to only active users of o1 to let them know they can turn on streaming if they’ve already integrated o1 into their applications.
4 Likes
For Vision, o1 is all I am using. For code I am using o3 in the API via Cursor / VS Code.
It depends on the task.
For general stuff, talking, web, I use 4o.
But I do empathize on the confusion.
Sam Altman had a recent X post mentioning GPT-5 will adjudicate internally, so there will be no “choosing” on our part.