Genie

Training compute

6.6×10²² FLOP

Parameters

10.7B

Published

Feb 23, 2024

Genie is an AI model developed by Google DeepMind (United States), first published in February 2024. It works in the video and games domain, on tasks such as video generation, image-to-video and text-to-video.

Training it took an estimated 6.6×10²² FLOP of compute (estimation method: reported). The model has 10,700,000,000 parameters. It was trained on roughly 942B datapoints. Training ran on 256 Google TPU v5p.

Access: Unreleased. Its weights are not openly released. Epoch AI rates the confidence of this record as confident.

Full record

Organization: Google DeepMind
Country of organization: United States
Domain: Video, Games
Task: Video generation, Image-to-video, Text-to-video
Training compute: 6.6×10²² FLOP
Compute estimation method: Reported
Parameters: 10,700,000,000
Dataset size: 942B
Training hardware: Google TPU v5p
Chips used: 256
Training power draw: 273.6 kW
Model accessibility: Unreleased
Open weights: No
Epoch confidence: Confident