K2-Horizon capability result attribution Authors and original publisher: K2 Team, Institute of Foundation Models, Mohamed bin Zayed University of Artificial Intelligence. Original work: K2-Horizon: A Frontier Fleet of Open Source Language Models Technical report version: arXiv:2611.21003v1, submitted 16 November 2026 https://arxiv.org/abs/2611.21003v1 https://arxiv.org/html/2611.21003v1 The arXiv record and rendered paper explicitly identify this publication as Creative Commons Attribution 4.0 International (CC BY 4.0). https://creativecommons.org/licenses/by/4.0/ https://creativecommons.org/licenses/by/4.0/legalcode Retrieved: 19 November 2026 HTML SHA-256: 0cb64470cb83d549b5571c5de5444a55016ce2632233b4245fc08954ec391241 Changes: llm4fun transcribed 11 scalar results for the authors' own exact post-trained K2-Horizon-0.9B, -3.7B and -7B checkpoints from Tables 34-36. Scores were not recalculated. Original setup descriptions identify sampling, tool/scaffold settings and missing details. Comparison-model columns, pre-training/Base results, Uno/diffusion results, graphics, and benchmark questions/prompts/answers are excluded. No author endorsement is implied. The source describes publisher evaluations using the AA Intelligence Index v4.1.1 setup. The selected facts are from the publisher's original paper, not extracted from the restricted Artificial Analysis result feed. The paper license is not permission to reuse a third party's comparison data. No such column is included. This attribution concerns the result publication. It does not establish rights to benchmark questions, model weights, software or trademarks.