GAIA-2

Training compute

7.6×10²² FLOP

Parameters

8.7B

Published

Mar 26, 2025

GAIA-2 is an AI model developed by Wayve (United Kingdom), first published in March 2025. It works in the video, vision, multimodal and language domain, on tasks such as self-driving car, video generation and instruction interpretation.

Training it took an estimated 7.6×10²² FLOP of compute (estimation method: operation counting). The model has 8,685,000,000 parameters. Training ran on NVIDIA H100 SXM5 80GB.

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

Full record

Organization: Wayve
Country of organization: United Kingdom
Domain: Video, Vision, Multimodal, Language
Task: Self-driving car, Video generation, Instruction interpretation
Training compute: 7.6×10²² FLOP
Compute estimation method: Operation counting
Parameters: 8,685,000,000
Training hardware: NVIDIA H100 SXM5 80GB
Model accessibility: Unreleased
Open weights: No
Epoch confidence: Confident