Visko moves its Orbis live video world model onto Google Cloud TPUs
Overview
Visko and Google Cloud engineers moved Visko's Orbis live inference pipeline onto a single Cloud TPU slice in 30 days, according to a Google Cloud post.
The pipeline combines video generation, super-resolution upscaling, and synchronized audio, and the post says the full multimodal pipeline was running in real time on TPUs within that period.
The post says the teams brought end-to-end generation down to a rendering speed of 1.07x playback. It credits Ulysses context parallelism, Pallas splash attention, and a rewrite of the audio stage from torchax to pure JAX, which the post says ran 5.8x faster.
Written by AI from the articles below · updated Oct 15, 1:06 PM ET
Check the sources:
Article timeline
The articles in this story. Times are ET.
- Google Cloud · AI & Machine LearningOfficialVisko streams its Orbis live world model on Google Cloud TPUs in 30 days
AIVisko worked with Google Cloud TPU engineers to move its Orbis live world model onto a single Cloud TPU slice in 30 days. The pipeline renders at 1.07x playback speed, using Ulysses context parallelism, Pallas splash attention, and a rewrite of the audio stage from torchax to pure JAX, which ran 5.8x faster.
Heat trend
Not enough continuous observations to show a trend yet.