# ***Cheers: Decoupling Patch Details from Semantic Representations Enables Unified Multimodal Comprehension and Generation*** Yichen Zhang^1*, [Da Peng](https://pengda02.github.io/)^2*, [Zonghao Guo](https://scholar.google.com/citations?user=h1I6LJcAAAAJ&hl=zh-CN)^1†, Zijian Zhang³, Xuesong Yang³, Tong Sun³, Shichu Sun³, Yidan Zhang³, Yanghao Li¹, Haiyan Zhao¹, Wang Xu¹, Qi Shi¹, Yangang Sun¹, Chi Chen¹, Shuo Wang¹, Yukun Yan¹, Xu Han¹, Qiang Ma¹, [Wei Ke](https://scholar.google.com/citations?hl=en&user=BENt-uEAAAAJ)², Liang Wang³, Zhiyuan Liu¹, Maosong Sun¹ ¹Tsinghua University, ²Xi'an Jiaotong University, ³University of Chinese Academy of Sciences \* Equal contribution † Corresponding author ## 🌟 What is ***Cheers***? A recent cutting-edge topic in multimodal modeling is to unify visual comprehension and generation within a single model. However, the two tasks demand mismatched decoding regimes and visual representations, making it non-trivial to jointly optimize within a shared feature space. In this work, we present ***Cheers***, a unified multimodal model that decouples patch-level details from semantic representations, thereby stabilizing semantics for multimodal understanding and improving fidelity for image generation via gated detail residuals. ***Cheers*** includes three key components: (i) a unified vision tokenizer that encodes and compresses image latent states into semantic tokens for efficient LLM conditioning, (ii) an LLM-based Transformer that unifies autoregressive decoding for text generation and diffusion decoding for image generation, and (iii) a cascaded flow matching head that decodes visual semantics first and then injects semantically gated detail residuals from the vision tokenizer to refine high-frequency content. Experiments on popular benchmarks demonstrate that ***Cheers*** matches or surpasses advanced UMMs in both visual understanding and generation. Notably, ***Cheers*** outperforms the Tar-1.5B on the popular benchmarks GenEval and MMBench, while requiring only 20% of the training cost, indicating effective and efficient (i.e., 4x token compression) unified multimodal modeling. - **Repository:** [[https://github.com/AI9Stars/Cheers](https://github.com/AI9Stars/Cheers)] - **Paper:** [[https://arxiv.org/abs/2603.12793](https://arxiv.org/abs/2603.12793)]