Kimi Linear: An Expressive, Efficient Attention Architecture

N/A

K IMI L INEAR : A N E XPRESSIVE , E FFICIENT ATTENTION A RCHITECTURE T ECHNICAL R EPORT OF K IMI L INEAR Kimi Team ‡ https://github.com/MoonshotAI/Kimi-Linear arXiv:2510.26692v2 [cs.CL] 1 Nov 2025…

saved by

• Lydia Nottingham
• Will Anderson

related reading

• Kimi Linear: An Expressive, Efficient Attention Architecturealphaxiv.org
• [2510.26692] Kimi Linear: An Expressive, Efficient Attention Architecturearxiv.org
• ali (@waterloo_intern) on Xx.com
• [2607.07953] Linear Attention Architectures: Mechanisms, Trade-offs, and Cross-Layer Routingarxiv.org
• DeltaNet Explained (Part I) | Songlin Yangsustcsonglin.github.io
• Linear Attention Fundamentals | Hailey Schoelkopfhaileyschoelkopf.github.io
• The Big LLM Architecture Comparisonmagazine.sebastianraschka.com
• 2502.11089arxiv.org
• Meta harness makes 10 times better Kimi K3 chipluoluo.ai
• Log-Linear Attentionarxiv.org
• How trillion-parameter AI models work and how to serve themyoutube.com
• Parallelizing Linear Transformers with the Delta Rule over Sequence Lengtharxiv.org