Latent Sparse Attention: Towards Subquadratic Ultra-Long Context Attention

A paper submitted to arXiv on 30 Oct 2026 proposes Latent Sparse Attention (LSA), an attention method that hierarchically clusters keys in a learned latent space so queries score clusters instead of every key, keeping accuracy at ultra-long context. On GLM-5.3-FLASH (QWEN-3.8-27B) at 10M tokens, the distilled variant D-LSA speeds up one attention layer by 22.7x (72.4x) in prefill and 4.6x (42.5x) in decode over its original sparse attention (full attention) while keeping similar accuracy across tasks, and the natively trained N-LSA matches full-attention accuracy with a 1.6x per-layer training speedup at 128k tokens. The authors position LSA as a way to make prefill, decode and training cheaper for agentic workflows that depend on context windows above 1M tokens.

Computer Science Artificial Intelligence Submitted on 30 Oct 2026 Title:Latent Sparse Attention: Towards Subquadratic Ultra-Long Context Attention View PDF http://arxiv.org/pdf/2610.42147v1 HTML experimental https://arxiv.org/html/2610.42147v1 Abstract:Agentic workflows are increasingly shifting towards long-horizon tasks, such as open-ended scientific work, where models benefit from long context windows. However, at ultra-long context above 1M tokens , full attention is prohibitively expensive. A promising approach is sparse attention, in which a dedicated indexer selects a subset of keys for the attention. Yet current methods of indexing either still have quadratic cost in context size or lose accuracy at ultra-long context scale. We address this with Latent Sparse Attention LSA , an efficient attention method that maintains accuracy at ultra-long context. In LSA, keys are hierarchically clustered in a learned latent space, and queries score the hierarchy of clusters instead of scoring every key, making prefill, decode and training faster at long context. We propose two variants, Distilled-LSA D-LSA , which can be applied to pre-trained models via distillation, and Native-LSA N-LSA , which is trained natively on the next-token loss alone. On GLM-5.3-FLASH QWEN-3.8-27B at 10M tokens, D-LSA speeds up one attention layer by 22.7x 72.4x in prefill and 4.6x 42.5x in decode over its original sparse attention full attention , while keeping similar accuracy across a variety of tasks. N-LSA matches the accuracy of full attention with a training speedup of 1.6x per layer at 128k tokens. LSA offers practical and efficient attention for models to benefit from ultra-long context. Additional Features References & Citations Loading... Bibliographic and Citation Tools Bibliographic Explorer What is the Explorer? https://info.arxiv.org/labs/showcase.html arxiv-bibliographic-explorer Connected Papers What is Connected Papers? https://www.connectedpapers.com/about Litmaps What is Litmaps? https://www.litmaps.co/ scite Smart Citations What are Smart Citations? https://www.scite.ai/ Code, Data and Media Associated with this Article alphaXiv What is alphaXiv? https://alphaxiv.org/ CatalyzeX Code Finder for Papers What is CatalyzeX? https://www.catalyzex.com DagsHub What is DagsHub? https://dagshub.com/ Gotit.pub What is GotitPub? http://gotit.pub/faq Hugging Face What is Huggingface? https://huggingface.co/huggingface ScienceCast What is ScienceCast? https://sciencecast.org/welcome Demos Recommenders and Search Tools Influence Flower What are Influence Flowers? https://influencemap.cmlab.dev/ CORE Recommender What is CORE? https://core.ac.uk/services/recommender 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? Learn more about arXivLabs https://info.arxiv.org/labs/index.html .