titleFastFrames: Video Generation at Interactive Latency
authorsGene Chou, Guandao Yang, Fujun Luan, Lars Mescheder, Wei Dong, Stephan Richter, Vladlen Koltun
date2026-11-17
venuearXiv (preprint, not peer reviewed)
idarXiv:2611.23226
affiliationApple
upvotes23 (Hugging Face Daily Papers, 2026-11-19)
linksabs pdf hf code
cite
relatedOpenVLHarness: Real World Visual Agentic Tool-Use with Multimodal Memory
It Takes Two to Match: Co-Evolving Generative Retriever with Reinforcement Learning
Depth Pro 2: Sharp and Temporally Consistent Metric Monocular Depth
MintAct: A Unified Visual Agent for Digital Environments
Decoding Looped Transformers Better for (Almost) Free
abstractAutoregressive video diffusion models generate frames based on new observations and inputs, making them a promising medium for interactive visual experiences. However, existing models are orders of magnitude too slow for fluid interaction. This paper introduces FastFrames, a family of fast autoregressive video diffusion models that achieves interactive latency without compromising quality. We profile the inference pipeline of state-of-the-art models and find that operations often treated as fixed implementation details, such as VAE decoding and KV cache updates, in fact account for over half of end-to-end inference latency. We challenge the assumption that these components are indispensable and replace them with lightweight alternatives. Our 1.3B variant runs at 36 ms per frame, or 28 responses per second (RPS), on a single B200 GPU while matching the generation quality of a state-of-the-art model that is nearly 8x slower. Our 14B variant runs at 45 ms per frame, or 22 RPS, on an 8x B200 GPU node and yields significantly better visual quality and realism. We show that FastFrames enables real-time interaction in applications such as camera-conditioned and physics-conditioned video generation.

bibtex

@misc{chou2026fastframes,
  title={FastFrames: Video Generation at Interactive Latency},
  author={Gene Chou and Guandao Yang and Fujun Luan and Lars Mescheder and Wei Dong and Stephan Richter and Vladlen Koltun},
  year={2026},
  eprint={2611.23226},
  archivePrefix={arXiv},
  url={https://arxiv.org/abs/2611.23226}
}