# UniFlow: A Unified Pixel Flow Tokenizer for Visual Understanding and Generation

Source: [UniFlow: A Unified Pixel Flow Tokenizer for Visual Understanding and Generation](<https://www.sota2.com/research/paper/68fa7c0a1e209131cea8c0b9>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-03-04T05:00:05.078Z

Authors: Zhengrong Yue, Haiyu Zhang, Xiangyu Zeng, Boyu Chen, Chenting Wang, Shaobin Zhuang, Lu Dong, Yi Wang, Limin Wang, Yali Wang

Published: 2025-10-12T12:50:23.000Z

## Abstract

Tokenizer is a crucial component for both visual understanding and generation. To advance toward the ultimate goal of universal modeling, recent research has focused on developing a unified tokenizer. However, existing tokenizers face a significant performance trade-off between understanding and generation, stemming from the inherent conflict between high-level semantic abstraction and low-level pixel reconstruction. To tackle this challenge, we propose a generic and unified tokenizer, namely UniFlow, by flexibly adapting any visual encoder with a concise reconstruction decoder. Specifically, we introduce layer-wise adaptive self-distillation applied to the well-pretrained visual encoders, which enables UniFlow to simultaneously inherit the strong semantic features for visual understanding and flexibly adapt to model fine-grained details for visual generation. Moreover, we propose a lightweight patch-wise pixel flow decoder, which efficiently achieves high-fidelity pixel reconstruction by modeling a conditional flow from the noisy state back to the patch-wise pixel domain. By leveraging the semantic features as visual conditions for the decoder, we effectively alleviate the training conflicts between understanding and generation. Furthermore, the patch-wise learning strategy simplifies the data distribution, thereby improving training efficiency. Extensive experiments across 13 challenging benchmarks spanning 7 widely studied visual understanding and generation tasks demonstrate that UniFlow achieves a win-win outcome. For instance, our 7B UniFlow-XL not only surpasses the 14B TokenFlow-XL by 6.05% on average understanding benchmarks, but also achieves a competitive results in both visual reconstruction and generation, surpassing UniTok by 0.15 in rFID and 0.09 in gFID (without guidance), respectively.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Image Classification on ImageNet-1K (Linear Probing/k-NN Evaluation)](<https://www.sota2.com/research/sota/image-classification-on-imagenet-1k-linear-probing-k-nn-evaluation.md>): Image Classification; ImageNet-1K; Linear Probing Accuracy: 82.6.

- [Image Reconstruction on MS-COCO 256x256 2017 (val)](<https://www.sota2.com/research/sota/image-reconstruction-on-ms-coco-256x256-2017-val.md>): Image Reconstruction; MS-COCO 256x256 2017 (val); PSNR: 32.48.

- [Multimodal Understanding on Multiple Datasets Aggregate](<https://www.sota2.com/research/sota/multimodal-understanding-on-multiple-datasets-aggregate.md>): Multimodal Understanding; Multiple Datasets Aggregate; Average Score: 79.09.

- [Image Reconstruction on ImageNet pre-training resolutions](<https://www.sota2.com/research/sota/image-reconstruction-on-imagenet-pre-training-resolutions.md>): Image Reconstruction; ImageNet pre-training resolutions; PSNR: 32.48.

- [Monocular Depth Estimation on NYU Depth v2 (test)](<https://www.sota2.com/research/sota/monocular-depth-estimation-on-nyu-depth-v2-test.md>): Monocular Depth Estimation; NYU Depth V2 (test); RMSE: 0.324.

- [Image Reconstruction on ImageNet-1K (full set)](<https://www.sota2.com/research/sota/image-reconstruction-on-imagenet-1k-full-set.md>): Image Reconstruction; ImageNet-1K (full set); PSNR: 32.32.

- [Image Reconstruction on MS-COCO 2017 (full set)](<https://www.sota2.com/research/sota/image-reconstruction-on-ms-coco-2017-full-set.md>): Image Reconstruction; MS-COCO 2017 (full set); PSNR: 32.29.

- [Visual Question Answering on GQA (Score)](<https://www.sota2.com/research/sota/visual-question-answering-on-gqa-score.md>): Visual Question Answering; GQA; Score: 65.86.

- [Image Reconstruction on ImageNet-1k 256x256 (val)](<https://www.sota2.com/research/sota/image-reconstruction-on-imagenet-1k-256x256-val.md>): Image Reconstruction; ImageNet-1k 256 x 256 (val); rFID: 0.26.

- [Multimodal Understanding on MME](<https://www.sota2.com/research/sota/multimodal-understanding-on-mme.md>): Multimodal Understanding; MME; MME Score: 2063.

- [Text-to-Image Generation on DPG-Bench (DPG Metric)](<https://www.sota2.com/research/sota/text-to-image-generation-on-dpg-bench-dpg-metric.md>): Text-to-Image Generation; DPG-Bench; DPG Score: 84.76.

- [Semantic Segmentation on ADE20K](<https://www.sota2.com/research/sota/semantic-segmentation-on-ade20k.md>): Semantic segmentation; ADE20K; mIoU: 55.4.

- [Text-to-Image Generation on GenEval (GenEval Metric)](<https://www.sota2.com/research/sota/text-to-image-generation-on-geneval-geneval-metric.md>): Text-to-Image Generation; GenEval; GenEval Score: 0.65.

- [Class-conditional image generation on ImageNet 256x256 (val)](<https://www.sota2.com/research/sota/class-conditional-image-generation-on-imagenet-256x256-val.md>): Class-conditional Image Generation; ImageNet 256x256 (val); Inception Score (IS): 290.

- [Object Hallucination Evaluation on POPE](<https://www.sota2.com/research/sota/object-hallucination-evaluation-on-pope.md>): Object Hallucination Evaluation; POPE; Accuracy: not recorded.

- [Multimodal Understanding on MMBench](<https://www.sota2.com/research/sota/multimodal-understanding-on-mmbench.md>): Multimodal Understanding; MMBench; Accuracy: not recorded.

- [Multi-discipline Multimodal Understanding on MMMU](<https://www.sota2.com/research/sota/multi-discipline-multimodal-understanding-on-mmmu.md>): Multi-discipline Multimodal Understanding; MMMU; Accuracy: not recorded.

- [Object Detection on MS-COCO 2017 (val)](<https://www.sota2.com/research/sota/object-detection-on-ms-coco-2017-val.md>): Object Detection; MS-COCO 2017 (val); mAP: not recorded.

- [Class-Conditional Image Generation on ImageNet 256x256 (train val)](<https://www.sota2.com/research/sota/class-conditional-image-generation-on-imagenet-256x256-train-val.md>): Class-conditional Image Generation; ImageNet 256x256 (train val); FID: not recorded.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2510.10575>)

[PDF](<https://arxiv.org/pdf/2510.10575.pdf>)
