Cosmos-Transfer1-7B

Training compute

3.6×10²⁴ FLOP

Parameters

7B

Published

Mar 18, 2025

Cosmos-Transfer1-7B is an AI model developed by NVIDIA (United States), first published in March 2025. It works in the video, vision and robotics domain, on tasks such as robotic manipulation, system control and video generation.

Training it took an estimated 3.6×10²⁴ FLOP of compute (estimation method: hardware). The model has 7,000,000,000 parameters. Training ran on 1,024 NVIDIA H100 SXM5 80GB for about 2K hours.

Access: Open weights (restricted use). Its weights are openly available. It is built on top of Cosmos-Predict1-7b-Video2World. Epoch AI rates the confidence of this record as likely.

Full record

Organization: NVIDIA
Country of organization: United States
Domain: Video, Vision, Robotics
Task: Robotic manipulation, System control, Video generation
Training compute: 3.6×10²⁴ FLOP
Compute estimation method: Hardware
Parameters: 7,000,000,000
Training hardware: NVIDIA H100 SXM5 80GB
Chips used: 1,024
Training time: 2,016 h
Training power draw: 1.4 MW
Model accessibility: Open weights (restricted use)
Open weights: Yes
Base model: Cosmos-Predict1-7b-Video2World
Epoch confidence: Likely