[2510.03215] Cache-to-Cache: Direct Semantic Communication Between Large Language Models (https://arxiv.org/static/browse/0.3.4/images/icons/apple-touch-icon.png) (../static/browse/0.3.4/images/icons/favicon-32x32.png) (../static/browse/0.3.4/images/icons/favicon-16x16.png) (../static/browse/0.3.4/images/icons/site.webmanifest) (https://arxiv.org/static/browse/0.3.4/images/icons/safari-pinned-tab.svg) (../static/browse/0.3.4/css/arXiv.css@v=20260318.css)  (../static/browse/0.3.4/css/arXiv-print.css@v=20200611.css)  (../static/browse/0.3.4/css/browse_search.css)  (../static/base/1.0.1/css/arxiv-header-footer.css@v=20260626.css)  (2510.03215.html)                  (../static/browse/0.3.4/css/tooltip.css)  (../../static.arxiv.org/js/bibex-dev/bibex.css@20200709.css)                (2510.03215.html#content) Skip to main content (https://arxiv.org/IgnoreMe)  (https://arxiv.org/) (archive)       (https://arxiv.org/search)    Search  (https://arxiv.org/user/create) Submit (https://info.arxiv.org/about/donate.html) Donate  (https://arxiv.org/login) Log in   Search arXiv (Search papers by title, author, abstract, or ID...) (all) (header)  Press Enter to search · (https://arxiv.org/search/advanced) Advanced search    Computer Science > Computation and Language  arXiv:2510.03215 (cs)  (../static/base/1.0.1/css/abs.css) [Submitted on 3 Oct 2025 ((https://arxiv.org/abs/2510.03215v1) v1 ), last revised 2 Mar 2026 (this version, v2)] Title: Cache-to-Cache: Direct Semantic Communication Between Large Language Models Authors: (https://arxiv.org/search/cs?searchtype=author&query=Fu,+T) Tianyu Fu , (https://arxiv.org/search/cs?searchtype=author&query=Min,+Z) Zihan Min , (https://arxiv.org/search/cs?searchtype=author&query=Zhang,+H) Hanling Zhang , (https://arxiv.org/search/cs?searchtype=author&query=Yan,+J) Jichao Yan , (https://arxiv.org/search/cs?searchtype=author&query=Dai,+G) Guohao Dai , (https://arxiv.org/search/cs?searchtype=author&query=Ouyang,+W) Wanli Ouyang , (https://arxiv.org/search/cs?searchtype=author&query=Wang,+Y) Yu Wang  View a PDF of the paper titled Cache-to-Cache: Direct Semantic Communication Between Large Language Models, by Tianyu Fu and 6 other authors (https://arxiv.org/pdf/2510.03215) View PDF (https://arxiv.org/html/2510.03215v2) HTML (experimental) Abstract: Multi-LLM systems harness the complementary strengths of diverse Large Language Models, achieving performance and efficiency gains that are not attainable by a single model. In existing designs, LLMs communicate through text, forcing internal representations to be transformed into output token sequences. This process both loses rich semantic information and incurs token-by-token generation latency. Motivated by these limitations, we ask: Can LLMs communicate beyond text? Oracle experiments show that enriching the KV-Cache semantics can improve response quality without increasing cache size, supporting KV-Cache as an effective medium for inter-model communication. Thus, we propose Cache-to-Cache (C2C), a new paradigm for direct semantic communication between LLMs. C2C uses a neural network to project and fuse the source model's KV-cache with that of the target model to enable direct semantic transfer. A learnable gating mechanism selects the target layers that benefit from cache communication. Compared with text communication, C2C utilizes the deep, specialized semantics from both models, while avoiding explicit intermediate text generation. Experiments show that C2C achieves 6.4-14.2% higher average accuracy than individual models. It further outperforms the text communication paradigm by approximately 3.1-5.4%, while delivering an average 2.5x speedup in latency. Our code is available at (https://github.com/thu-nics/C2C) this https URL .  Comments: Published in ICLR'26  Subjects: Computation and Language (cs.CL) ; Machine Learning (cs.LG)  (Mathematical Subject Classification) MSC classes: 68T07, 68T50  (Association of Computing Machinery Classification) ACM classes: I.2.7  Cite as: (2510.03215.html) arXiv:2510.03215 [cs.CL]    (or (https://arxiv.org/abs/2510.03215v2) arXiv:2510.03215v2 [cs.CL] for this version)    (https://doi.org/10.48550/arXiv.2510.03215) https://doi.org/10.48550/arXiv.2510.03215   Focus to learn more   arXiv-issued DOI via DataCite        Submission history From: Tianyu Fu [(https://arxiv.org/show-email/5bc6748f/2510.03215) view email ]  (https://arxiv.org/abs/2510.03215v1) [v1]  Fri, 3 Oct 2025 17:52:32 UTC (484 KB) [v2] Mon, 2 Mar 2026 19:24:02 UTC (546 KB)    Full-text links: Access Paper: View a PDF of the paper titled Cache-to-Cache: Direct Semantic Communication Between Large Language Models, by Tianyu Fu and 6 other authors (https://arxiv.org/pdf/2510.03215) View PDF  (https://arxiv.org/html/2510.03215v2) HTML (experimental)  (https://arxiv.org/src/2510.03215) TeX Source    (http://arxiv.org/licenses/nonexclusive-distrib/1.0/) (Rights to this article) view license   Current browse context: cs.CL (https://arxiv.org/prevnext?id=2510.03215&function=prev&context=cs.CL) (previous in cs.CL (accesskey p)) < prev  |  (https://arxiv.org/prevnext?id=2510.03215&function=next&context=cs.CL) (next in cs.CL (accesskey n)) next >    (https://arxiv.org/list/cs.CL/new) new |  (https://arxiv.org/list/cs.CL/recent) recent |  (https://arxiv.org/list/cs.CL/2025-10) 2025-10  Change to browse by: (https://arxiv.org/abs/2510.03215?context=cs) cs (https://arxiv.org/abs/2510.03215?context=cs.LG) cs.LG    References & Citations (https://ui.adsabs.harvard.edu/abs/arXiv:2510.03215) NASA ADS  (https://scholar.google.com/scholar_lookup?arxiv_id=2510.03215) Google Scholar  (https://api.semanticscholar.org/arXiv:2510.03215) Semantic Scholar     export BibTeX citation Loading...  BibTeX formatted citation ×  loading...  Data provided by:      Bookmark  (http://www.bibsonomy.org/BibtexHandler?requTask=upload&url=https://arxiv.org/abs/2510.03215&description=Cache-to-Cache: Direct Semantic Communication Between Large Language Models) (Bookmark on BibSonomy) (BibSonomy)   (https://reddit.com/submit?url=https://arxiv.org/abs/2510.03215&title=Cache-to-Cache: Direct Semantic Communication Between Large Language Models) (Bookmark on Reddit) (Reddit)     Bibliographic Tools Bibliographic and Citation Tools  Bibliographic Explorer Toggle   Bibliographic Explorer ((https://info.arxiv.org/labs/showcase.html#arxiv-bibliographic-explorer) What is the Explorer? )    Connected Papers Toggle   Connected Papers ((https://www.connectedpapers.com/about) What is Connected Papers? )    Litmaps Toggle   Litmaps ((https://www.litmaps.co/) What is Litmaps? )    scite.ai Toggle   scite Smart Citations ((https://www.scite.ai/) What are Smart Citations? )         Code, Data, Media Code, Data and Media Associated with this Article  alphaXiv Toggle   alphaXiv ((https://alphaxiv.org/) What is alphaXiv? )    Links to Code Toggle   CatalyzeX Code Finder for Papers ((https://www.catalyzex.com) What is CatalyzeX? )    DagsHub Toggle   DagsHub ((https://dagshub.com/) What is DagsHub? )    GotitPub Toggle   Gotit.pub ((http://gotit.pub/faq) What is GotitPub? )    Huggingface Toggle   Hugging Face ((https://huggingface.co/huggingface) What is Huggingface? )    ScienceCast Toggle   ScienceCast ((https://sciencecast.org/welcome) What is ScienceCast? )           Demos Demos  Replicate Toggle   Replicate ((https://replicate.com/docs/arxiv/about) What is Replicate? )    Spaces Toggle   Hugging Face Spaces ((https://huggingface.co/docs/hub/spaces) What is Spaces? )    Spaces Toggle   TXYZ.AI ((https://txyz.ai) What is TXYZ.AI? )        Related Papers Recommenders and Search Tools  Link to Influence Flower   Influence Flower ((https://influencemap.cmlab.dev/) What are Influence Flowers? )    Core recommender toggle   CORE Recommender ((https://core.ac.uk/services/recommender) What is CORE? )     Author  Venue  Institution  Topic                About arXivLabs  arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? (https://info.arxiv.org/labs/index.html) Learn more about arXivLabs  .                (https://arxiv.org/auth/show-endorsers/2510.03215) Which authors of this paper are endorsers? | Disable MathJax ((https://info.arxiv.org/help/mathjax.html) What is MathJax? )      We gratefully acknowledge support from
        our major funders , (https://info.arxiv.org/about/ourmembers.html) member institutions  ,   ,
        and all contributors.  (https://info.arxiv.org/about) About · (https://info.arxiv.org/help) Help · (https://info.arxiv.org/help/contact.html) Contact · (https://info.arxiv.org/help/subscribe) Subscribe · (https://info.arxiv.org/help/license/index.html) Copyright · (https://info.arxiv.org/help/policies/privacy_policy.html) Privacy · (https://info.arxiv.org/help/web_accessibility.html) Accessibility · (https://status.arxiv.org) Operational Status(opens in new tab)    Major funding support from (https://www.simonsfoundation.org/) (Simons Foundation)  (https://www.sfi.org.bm/) (Simons Foundation International)  (https://www.schmidtsciences.org/) (Schmidt Sciences)