Wan 2.2 image-to-video anime characters having a lightsaber fight. I noticed it’s not as fluid as MidJourney Video, but honestly it seems really fucking good.

At first it was performative but now I really do have a problem

They didnt see the ships

You can use "think longer" with gpt4o now on ChatGPT but apparently its just routing to o3

I hate that labs can fuck with the quant of their models without telling you, making the model worse, and still charge you the same price.

This explains why I am on twitter.

“gpt-5-alpha” well fuck me looks like we are getting the API day one

My agent worked 56 minutes to create a "Welcome to earth" powerpoint presentation for GPT-5

Be not afraid

Funny how if you add “gpt-5-alpha” as a custom model in Cursor, it automatically recognizes it as an official model with the “brain” symbol.

This update made Cursor fucking unusable. Why can’t I give the full context of code files to the model anymore?

Only subscribe to Cursor if you hate yourself. They will find 7 new ways to fuck you every update. They want you crying, begging to be treated like a human being.