Vision & Multimodal Sep 2025 · #67 most cited · 82 citations

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

Junsong Chen et al.

arXiv:2509.24695

In short

SANA-Video is a small diffusion model for high-resolution, minute-long video built on linear attention, with a constant-memory cache for block-by-block generation. It trained for about 1% of MovieGen’s cost and is 16× faster than comparable small models.

Why it matters

Quality video generation that fits on a single consumer GPU.

Read first

The 3 Field Guide ideas this paper leans on.

Vision & Multimodal · Landmark Diffusion Model A generative model that learns to denoise data, achieving state-of-the-art image generation (Stable Diffusion, DALL-E 2). Language & LLMs · Landmark Self-Attention A mechanism where each token attends to all other tokens in the sequence to understand contextual relationships. Shipping AI · Standard Inference Latency The time delay between submitting input and receiving output from a deployed model, critical for real-time applications.

Starting from scratch? The full route 13 ideas · basics first

Machine Learning ✓ understood
Building systems that learn patterns from data instead of following hand-written rules, getting better at a task as they see more examples.
Neural Network ✓ understood
A computational model inspired by biological neural networks, consisting of interconnected nodes (neurons) organized in layers that process information through weighted connections.
Diffusion Model · read first ✓ understood
A generative model that learns to denoise data, achieving state-of-the-art image generation (Stable Diffusion, DALL-E 2).
Dataset ✓ understood
A collection of data examples used for training, validating, or testing machine learning models.
Feature ✓ understood
A single measurable property of an example, such as a house's floor area or how many links an email contains, used as an input to a model.
Deep Learning ✓ understood
A subset of machine learning that uses neural networks with multiple layers (deep neural networks) to learn hierarchical representations of data.
Representation Learning ✓ understood
Learning useful features or representations of data automatically, rather than hand-crafting them.
Embedding ✓ understood
A list of numbers (a vector) that represents a word, sentence, image or other item, learned so that similar items end up close together.
Attention Mechanism ✓ understood
A technique that lets a neural network weigh every part of its input when producing each output, focusing on the parts most relevant at that step.
Self-Attention · read first ✓ understood
A mechanism where each token attends to all other tokens in the sequence to understand contextual relationships.
Training ✓ understood
The process of fitting a model to data by repeatedly measuring how wrong its outputs are and adjusting its parameters to reduce that error.
Inference ✓ understood
Running a trained model on new inputs to get predictions, with its weights frozen: the stage of a model's life that users actually interact with.
Inference Latency · read first ✓ understood
The time delay between submitting input and receiving output from a deployed model, critical for real-time applications.

In the frontier

Rank: #67 of 100
Citations: 82; as of Aug 9, 2026
Published: Sep 2025

Topics: Video generation and world models , Model architecture , Efficiency and serving

Selection: 1kpapers.com by Together AI, most-cited as of Aug 9, 2026

Nearby papers

Canon · 2017 Attention Is All You Need Every major LLM, and most modern vision and speech models, is a transformer. Canon · 2020 Denoising Diffusion Probabilistic Models DDPM is the recipe behind Stable Diffusion, DALL·E, Midjourney and today’s video models. Canon · 2021 High-Resolution Image Synthesis with Latent Diffusion Models It’s the architecture behind Stable Diffusion, which put high-quality text-to-image generation on ordinary GPUs. Canon · 2022 FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness It is a big part of why long context windows became affordable, and it ships inside nearly every LLM stack.