Most people do not realize how difficult it is to bring up new AI training compute clusters. The amount of issues that come up with bringing up new GPUs and getting them ready for large-scale synchronous training are unfathomable. On top of the hardware issues, a whole new set of

.Grok new office decor 🦅

.xAI has the best food. Come join us

This is a pretty neat paper that shows the reasoning capability of video diffusion models. Our CVPR 2025 work shows similar results on how to take pre-trained generalist image diffusion models and apply them for perceptual tasks like depth estimation and occlusion prediction. I

The people you work and collaborate with silently affect your life trajectory. Many times people will adopt a mentor or peer's interest before testing their own. It's important early on to reflect on your influences to see if they're hindering the development of your own

Ten years ago, it was exciting to see AI that can recognize different types of cats, and today, we have AI systems that can generate new realities within seconds. In a few years, it’ll be normalized for most of the digital content that people consume to be AI generated. This is

We’ve just updated our @Grok Imagine video generation model, check it out and let us know what new features you’d like to see! It’s been really fun to work with the team to train this model 🚀

create the future with Grok imagine

Check out the new @Grok Imagine v0.9 model! Our team spent the last month working extremely hard to deliver this massive upgrade. The iteration speed here @xAI is absolutely ludicrous 🚀

create new stories and realities with Grok imagine x.com/amXFreeze/stat…

Our @Grok Imagine model can also simulate basic robot manipulation tasks without any robotics specific training. It’s only going to get better from here!

thanks for the tungsten cube @Grok :)

really great and exciting work!!

𝕏 works

another night, another yolo 👨‍🍳

pretty cool

building at ludicrous speed @xAI 🏎️💨

Emergent zero-shot reasoning with @Grok Imagine