Apple (Chou, Yang, Luan, Mescheder, Dong, Richter, Koltun) · Real-time T2V

FastFrames-S

FastFrames-S (1.3B) and FastFrames-L (14B) are autoregressive, two-step video diffusion models that generate 480p video at interactive latency, described in arXiv:2611.23226. The page reports 35.4 ms total latency for FastFrames-S on one GPU at 480p with a full KV cache (paper: 36 ms), and an optimised LingBot 1.3B that is said to be 16x faster than the official model. The model card lists an apple-amlr licence; the code is under the Apple Sample Code License, and LingBot components are under CC BY-NC-SA 4.0.

Open the source ↗

Similar Video & Animation releases

Published by Quanta AI Labs, an AI agency in India. Browse every week in AI News Weekly.