Paper 2509.24695
SANA-Video: Efficient Video Generation with Block Linear Diffusion Transformer
- Published
- Sep 2025
- Research lab
- NVIDIA
- Citations
- 82
- GitHub
- 8.7K stars
01 In brief
Summary
SANA-Video is a small diffusion model for efficient, high-resolution (up to 720×1280) and minute-long video generation, deployable on RTX 5090 GPUs.
It uses a Linear DiT with linear attention (O(N) complexity) and a constant-memory KV cache for block linear attention, enabling long videos with fixed memory.
Training cost is 12 days on 64 H100 GPUs (1% of MovieGen).
It achieves competitive performance with state-of-the-art small models (e.g., Wan 2.1-1.3B) while being 16× faster.
Key designs include 3D RoPE in linear attention, temporal convolution in Mix-FFN, a deep compression video autoencoder (DCAE-V, F32T4C32), and a data filtering pipeline.
LongSANA variant generates 1-minute 480P videos in 35 seconds on H100.
On RTX 5090 with NVFP4 quantization, a 5-second 720p video is generated in 29 seconds (2.4× speedup).
The model supports T2I, T2V, and I2V tasks.
02 From the paper
Abstract
We introduce SANA-Video, a small diffusion model that can efficiently generate videos up to 720x1280 resolution and minute-length duration. SANA-Video synthesizes high-resolution, high-quality and long videos with strong text-video alignment at a remarkably fast speed, deployable on RTX 5090 GPU. Two core designs ensure our efficient, effective and long video generation: (1) Linear DiT: We leverage linear attention as the core operation, which is more efficient than vanilla attention given the large number of tokens processed in video generation. (2) Constant-Memory KV cache for Block Linear Attention: we design block-wise autoregressive approach for long video generation by employing a constant-memory state, derived from the cumulative properties of linear attention. This KV cache provides the Linear DiT with 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 of MovieGen. Given its low cost, SANA-Video achieves competitive performance compared to modern state-of-the-art small diffusion 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 with NVFP4 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.