[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-84375-en":3,"doc-seo-84375-105":29,"detail-sidebar-cat-0-en-105":83},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":13,"seo_description":14,"update_tm":27,"read_time":28},84375,13056703020460,"Valentina","https://ap-avatar.wpscdn.com/avatar/be000253dac470eee5d?_k=1778207105932848923",8,"Research & Report","DeltaV: Thinking with Visual State Updates in Unified Large Multimodal Models","Unified Large Multimodal Models support interleaved multimodal reasoning with intermediate visual states, yet they typically generate each state as a full image, causing visual-token redundancy and weak supervision on the sparse, reasoning-critical transitions. DeltaV replaces full-image generation with conditioned visual updates, incrementally predicting compact update tokens that represent changes across reasoning steps. A TSIM Router stops token allocation when marginal reconstruction gain is low. DeltaV uses StructCoT, a 1.05M-sample, 44-domain dataset, reducing new visual tokens by 55.6% and improving multimodal reasoning by 3.3%.","arXiv :2607 .08434v 1 [ cs .CV] 9 Jul 2026  \nDeltaV: Thinking with Visual State Updates in Unified  \nLarge Multimodal Models  \nPengjie Wang1,∗,‡, Linger Deng1,∗,‡, Zujia Zhang1 , Shaojie Zhang2 , Zhenbo Luo2 , Pei Fu2 ,  \nJian Luan2 , Xiang Bai1 , Yuliang Liu1,†  \n1 Huazhong University of Science and Technology 2 MiLM Plus, Xiaomi Inc.  \nAbstract  \nCurrent Unified Large Multimodal Models (ULMMs) support interleaved multimodal reasoning through textual reasoning and intermediate visual states, but typically generate each visual state as a full image. This full-image generation paradigm introduces substantial visual-token redundancy and dilutes supervision on sparse yet reasoning-critical state transitions. We propose DeltaV, a ULMM that replaces full-image generation with visual updates. Conditioned on historical visual states, DeltaV incrementally predicts compact update tokens that capture the visual changes across reasoning steps, avoiding repeated modeling of unchanged content. To align the token budget of each update with the magnitude of visual change, DeltaV introduces a temporal similarity (TSIM) Router, which stops allocating tokens once the marginal reconstruction gain falls below a threshold. To support more diverse and generalizable reasoning, we further construct StructCoT, a large-scale interleaved multimodal reasoning dataset with 1.05M samples spanning 44 task domains. Experiments show that the visual-update paradigm reduces newly generated visual tokens by 55.6% on average without compromising reconstruction fidelity, and improves multimodal reasoning by 3.3% over full-image generation. Trained with StructCoT and large-scale multimodal data, DeltaV-2B further outperforms substantially larger open-source models by 8.4% on in-domain multimodal reasoning evaluations and surpasses the comparable-scale Qwen3-VL-2Bby 5.9% on external multimodal reasoning and understanding benchmarks. Code, models, and StructCoT will be released at [https://github. com/Pengjie-W/DeltaV](https://github. com/Pengjie-W/DeltaV).  \n1 Introduction  \nMultimodal Large Language Models (MLLMs) (Bai et al., 2025a; Deng et al., 2026; Li et al., 2024; Wang et al., 2025b; Wu et al., 2024) have demonstrated strong performance on general vision-language benchmarks (Chen et al., 2024a; Liu et al., 2024) . However, they still struggle with complex tasks, such as embodied intelligence, scientific problem solving, and spatial intelligence, that require multi-step reasoning over evolving visual structures (Hao et al., 2025; Jiang et al., 2025) . To address this limitation, prior studies introduce intermediate visual states into the reasoning process, forming a think-and-sketch paradigm known as interleaved multimodal reasoning (Li et al., 2025b; Zheng et al., 2025) .  \n∗ Equal contribution.  \n‡Work done during internships at Xiaomi Inc.  \nBuried critical visual token  \nState 0 State 1  \nVisual token for full image  \nState 2 State 3  \n Error region Newly generated visual update token  \nState 1 State 2 State 3  \n(a) Multimodal Reasoning Input (b) Existing ULMMs: Reasoning through Complete Visual States (c) DeltaV: Reasoning through Visual State Transitions  \nFigure 1 From generating complete visual states to modeling how visual states evolve. Comparison between existing ULMMs and DeltaV for interleaved multimodal reasoning. Given the initial visual state tokens 􀀯0 tokenized from the input image, existing ULMMs generate each intermediate visual state as a full image, whereas DeltaV thinks with variation-aware visual updates and dynamically allocates tokens to critical visual changes.  \nUnified Large Multimodal Models (ULMMs) provide a promising end-to-end framework for interleaved multimodal reasoning (Chen et al., 2025; Deng et al., 2025), as they unify language reasoning and visual generation within a single model. While their native visual generation capability is impressive, its potential for supporting reasoning remains underexplored (Zou et al., 2025) ","cbCaiadehA9zhIiK","https://ap.wps.com/l/cbCaiadehA9zhIiK","pdf",29026789,1,36,"English","en",105,"# Abstract\n# Introduction","[{\"question\":\"What is StructCoT and what role does it play?\",\"answer\":\"StructCoT is a large-scale interleaved multimodal reasoning dataset with 1.05M samples across 44 task domains, used to train DeltaV and improve reasoning performance.\"}]",1784195180,91,{"code":4,"msg":30,"data":31},"ok",{"site_id":24,"language":23,"slug":32,"title":13,"keywords":33,"description":14,"schema_data":34,"social_meta":78,"head_meta":80,"extra_data":82,"updated_unix":27},"deltav-thinking-with-visual-state-updates-in-unified-large-multimodal-models","",{"@graph":35,"@context":77},[36,53,68],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,47,50],{"item":40,"name":41,"@type":42,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":46},"https://docshare.wps.com/document/","Document",2,{"item":48,"name":12,"@type":42,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/deltav-thinking-with-visual-state-updates-in-unified-large-multimodal-models/84375/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":23,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":40,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-17","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71],{"name":72,"@type":73,"acceptedAnswer":74},"What is StructCoT and what role does it play?","Question",{"text":75,"@type":76},"StructCoT is a large-scale interleaved multimodal reasoning dataset with 1.05M samples across 44 task domains, used to train DeltaV and improve reasoning performance.","Answer","https://schema.org",{"og:url":51,"og:type":79,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":81,"canonical":51},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":84},[85,89,93,97,102,107,112,115,120,123,127],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":86,"show_sort_weight":87,"slug":88},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":90,"show_sort_weight":91,"slug":92},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":94,"show_sort_weight":95,"slug":96},"Exam",70,"exam",{"id":98,"doc_module":4,"doc_module_name":45,"category_name":99,"show_sort_weight":100,"slug":101},5,"Comic",60,"comic",{"id":103,"doc_module":4,"doc_module_name":45,"category_name":104,"show_sort_weight":105,"slug":106},6,"Technology",50,"technology",{"id":108,"doc_module":4,"doc_module_name":45,"category_name":109,"show_sort_weight":110,"slug":111},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":113,"slug":114},30,"research-report",{"id":116,"doc_module":4,"doc_module_name":45,"category_name":117,"show_sort_weight":118,"slug":119},9,"Religion & Spirituality",20,"religion-spirituality",{"id":118,"doc_module":4,"doc_module_name":45,"category_name":121,"show_sort_weight":118,"slug":122},"World Cup","world-cup",{"id":124,"doc_module":4,"doc_module_name":45,"category_name":125,"show_sort_weight":124,"slug":126},10,"Lifestyle","lifestyle",{"id":128,"doc_module":4,"doc_module_name":45,"category_name":129,"show_sort_weight":98,"slug":130},19,"General","general"]