{"slug": "kimi-linear-an-expressive-efficient-attention-architecture", "title": "Kimi Linear: An Expressive, Efficient Attention Architecture", "summary": "Researchers at Moonshot AI introduced Kimi Linear, a hybrid linear attention architecture that outperforms full attention across short-context, long-context, and reinforcement learning scaling regimes. The architecture's core, Kimi Delta Attention (KDA), extends Gated DeltaNet with a finer-grained gating mechanism and achieves up to 75% KV cache reduction and 6x decoding throughput for 1M context. The team open-sourced the KDA kernel, vLLM implementations, and model checkpoints.", "body_md": "# Computer Science > Computation and Language\n\n[Submitted on 30 Oct 2025 (\n\n[v1](https://arxiv.org/abs/2510.26692v1)), last revised 1 Nov 2025 (this version, v2)]# Title:Kimi Linear: An Expressive, Efficient Attention Architecture\n\n[View PDF](/pdf/2510.26692)\n\nAbstract:We introduce Kimi Linear, a hybrid linear attention architecture that, for the first time, outperforms full attention under fair comparisons across various scenarios -- including short-context, long-context, and reinforcement learning (RL) scaling regimes. At its core lies Kimi Delta Attention (KDA), an expressive linear attention module that extends Gated DeltaNet with a finer-grained gating mechanism, enabling more effective use of limited finite-state RNN memory. Our bespoke chunkwise algorithm achieves high hardware efficiency through a specialized variant of the Diagonal-Plus-Low-Rank (DPLR) transition matrices, which substantially reduces computation compared to the general DPLR formulation while remaining more consistent with the classical delta rule.\n\nWe pretrain a Kimi Linear model with 3B activated parameters and 48B total parameters, based on a layerwise hybrid of KDA and Multi-Head Latent Attention (MLA). Our experiments show that with an identical training recipe, Kimi Linear outperforms full MLA with a sizeable margin across all evaluated tasks, while reducing KV cache usage by up to 75% and achieving up to 6 times decoding throughput for a 1M context. These results demonstrate that Kimi Linear can be a drop-in replacement for full attention architectures with superior performance and efficiency, including tasks with longer input and output lengths.\n\nTo support further research, we open-source the KDA kernel and vLLM implementations, and release the pre-trained and instruction-tuned model checkpoints.\n\n## Submission history\n\nFrom: Yulun Du [[view email](/show-email/19b5afec/2510.26692)]\n\n**Thu, 30 Oct 2025 16:59:43 UTC (645 KB)**\n\n[[v1]](/abs/2510.26692v1)**[v2]** Sat, 1 Nov 2025 12:05:18 UTC (691 KB)\n\n### References & Citations\n\nLoading...\n\n# Bibliographic and Citation Tools\n\nBibliographic Explorer\n\n*(*[What is the Explorer?](https://info.arxiv.org/labs/showcase.html#arxiv-bibliographic-explorer))\nConnected Papers\n\n*(*[What is Connected Papers?](https://www.connectedpapers.com/about))\nLitmaps\n\n*(*[What is Litmaps?](https://www.litmaps.co/))\nscite Smart Citations\n\n*(*[What are Smart Citations?](https://www.scite.ai/))# Code, Data and Media Associated with this Article\n\nalphaXiv\n\n*(*[What is alphaXiv?](https://alphaxiv.org/))\nCatalyzeX Code Finder for Papers\n\n*(*[What is CatalyzeX?](https://www.catalyzex.com))\nDagsHub\n\n*(*[What is DagsHub?](https://dagshub.com/))\nGotit.pub\n\n*(*[What is GotitPub?](http://gotit.pub/faq))\nHugging Face\n\n*(*[What is Huggingface?](https://huggingface.co/huggingface))\nScienceCast\n\n*(*[What is ScienceCast?](https://sciencecast.org/welcome))# Demos\n\n# Recommenders and Search Tools\n\nInfluence Flower\n\n*(*[What are Influence Flowers?](https://influencemap.cmlab.dev/))\nCORE Recommender\n\n*(*[What is CORE?](https://core.ac.uk/services/recommender))# arXivLabs: experimental projects with community collaborators\n\narXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.\n\nBoth 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.\n\nHave an idea for a project that will add value for arXiv's community? [ Learn more about arXivLabs](https://info.arxiv.org/labs/index.html).", "url": "https://wpnews.pro/news/kimi-linear-an-expressive-efficient-attention-architecture", "canonical_source": "https://arxiv.org/abs/2510.26692", "published_at": "2026-07-21 02:49:28+00:00", "updated_at": "2026-07-21 03:23:30.485700+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-research", "ai-infrastructure"], "entities": ["Moonshot AI", "Kimi Linear", "Kimi Delta Attention", "Gated DeltaNet", "Multi-Head Latent Attention", "vLLM"], "alternates": {"html": "https://wpnews.pro/news/kimi-linear-an-expressive-efficient-attention-architecture", "markdown": "https://wpnews.pro/news/kimi-linear-an-expressive-efficient-attention-architecture.md", "text": "https://wpnews.pro/news/kimi-linear-an-expressive-efficient-attention-architecture.txt", "jsonld": "https://wpnews.pro/news/kimi-linear-an-expressive-efficient-attention-architecture.jsonld"}}