Paper page - SANA-Video: Efficient Video Generation with Block Linear Diffusion Transformer

Source: https://huggingface.co/papers/2509.24695 Published on Sep 29, 2025

·

Submitted byhttps://huggingface.co/Yuyang-z

Yuyangon Sep 30, 2025

Authors:

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

Abstract

SANA-Video, a small diffusion model, efficiently generates high-resolution, high-quality videos with strong text-video alignment using linear attention and a constant-memory KV cache, achieving competitive performance at a lower cost and faster speed.

We introduce SANA-Video, a smalldiffusion modelthat can efficiently generate videos up to 720x1280 resolution and minute-length duration. SANA-Video synthesizes high-resolution, high-quality and long videos with strongtext-video alignmentat a remarkably fast speed, deployable on RTX 5090 GPU. Two core designs ensure our efficient, effective and long video generation: (1)Linear DiT: We leveragelinear attentionas the core operation, which is more efficient than vanilla attention given the large number of tokens processed in video generation. (2)Constant-Memory KV cachefor Block Linear Attention: we designblock-wise autoregressiveapproach for long video generation by employing a constant-memory state, derived from the cumulative properties oflinear attention. This KV cache provides theLinear DiTwith global context at a fixed memory cost, eliminating the need for a traditional KV cache and enabling efficient, minute-long video generation. In addition, we explore effective data filters and model training strategies, narrowing the training cost to 12 days on 64 H100 GPUs, which is only 1% of the cost ofMovieGen. Given its low cost, SANA-Video achieves competitive performance compared to modern state-of-the-art smalldiffusion models (e.g., Wan 2.1-1.3B and SkyReel-V2-1.3B) while being 16x faster in measured latency. Moreover, SANA-Video can be deployed on RTX 5090 GPUs withNVFP4 precision, accelerating the inference speed of generating a 5-second 720p video from 71s to 29s (2.4x speedup). In summary, SANA-Video enables low-cost, high-quality video generation.

View arXiv pageView PDFProject pageGitHub5.25kAdd to collection

Get this paper in your agent:

hf papers read 2509\.24695

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper5

#### Efficient-Large-Model/SANA-Video_2B_720p Text-to-Video• UpdatedMar 16 • 20 • 15 #### Efficient-Large-Model/SANA-Video_2B_480p Text-to-Video• UpdatedOct 28, 2025 • 88 • 14 #### Efficient-Large-Model/SANA-Video_2B_480p_diffusers Text-to-Video• UpdatedNov 4, 2025 • 5 #### Efficient-Large-Model/SANA-Video_2B_480p_LongLive_diffusers Text-to-Video• UpdatedDec 9, 2025 • 2 Browse 5 models citing this paper## Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2509.24695 in a dataset README.md to link it from this page.

Spaces citing this paper1

Collections including this paper5

Browse 5 collections that include this paper