The Year in AI Papers
The year/Independent research

Paper 2601.02204

NextFlow: Unified Sequential Modeling Activates Multimodal Understanding and Generation

Huichao Zhang, Liao Qu, Yiheng Liu, Hang Chen, Yangyang Song, Yongsheng Dong, Shikun Sun, Xian Li, Xu Wang, Yi Jiang, Hu Ye, Bo Chen, et al.

Published
Jan 2026
Research lab
Independent
Citations
12
GitHub
331 stars

01 In brief

Summary

NextFlow is a unified decoder-only autoregressive transformer trained on 6 trillion interleaved text-image tokens.

It uses a dual-codebook tokenizer for semantic and pixel-level features, and adopts next-scale prediction for visual generation instead of raster-scan, enabling 1024x1024 image generation in 5 seconds.

The model retains next-token prediction for text and uses a multi-scale 3D RoPE for positional encoding.

Training includes a progressive resolution curriculum (256, 512, 1024), scale reweighting, self-correction with residual features, and a prefix-tuning GRPO strategy for reinforcement learning.

An optional diffusion decoder refines details.

NextFlow achieves state-of-the-art performance among unified models, rivaling diffusion baselines in visual quality, and excels in image editing, interleaved generation, and chain-of-thought reasoning.

It requires 6x fewer FLOPs than MMDiT-based diffusion models at 1024 resolution.

02 From the paper

Abstract

We present NextFlow, a unified decoder-only autoregressive transformer trained on 6 trillion interleaved text-image discrete tokens. By leveraging a unified vision representation within a unified autoregressive architecture, NextFlow natively activates multimodal understanding and generation capabilities, unlocking abilities of image editing, interleaved content and video generation. Motivated by the distinct nature of modalities - where text is strictly sequential and images are inherently hierarchical - we retain next-token prediction for text but adopt next-scale prediction for visual generation. This departs from traditional raster-scan methods, enabling the generation of 1024x1024 images in just 5 seconds - orders of magnitude faster than comparable AR models. We address the instabilities of multi-scale generation through a robust training recipe. Furthermore, we introduce a prefix-tuning strategy for reinforcement learning. Experiments demonstrate that NextFlow achieves state-of-the-art performance among unified models and rivals specialized diffusion baselines in visual quality.