Vibeleaderboard
← All Intel
Intel / video

Inside xAI: Building Grok Imagine in 3 Months, Videogen vs World Models, and Video Agents— Ethan He

Source
youtube.com
Author
Latent Space
Date

Context

Ethan He, who worked on NVIDIA's Cosmos video before joining xAI in May 2025, says the team shipped the first Grok Imagine video model in about three months. In a Latent Space interview, he describes the starting point as a few engineers with no infrastructure, data or model, and attributes the speed to a small, close team and to xAI's existing data and infrastructure.

He argues the key measure is how many experiment iterations a team can run per day, since each one means preparing data, training a model and evaluating it against the previous version. As a general , he adds that many improvements come from finding small bugs in data and training pipelines rather than from new algorithms. These are his recollections. He says he cannot comment on specifics of how the model was built, and the excerpt does not cover the model's results.

Terms in this piece · Glossary
  • Model — A computational system that uses learned patterns to produce an output from information supplied to it.
  • Training — The process that adjusts a model using examples and a measure of how well it performs.
  • Observation — The result the harness returns into the context after executing a tool: contents, a listing, a status or an error.
Read the source www.youtube.com
More from Latent Space
Recommended reads
Comments

Checking sign-in…

Loading comments…