Daily AI Papers — September 15, 2026

1. Atria Dawn: The Dawn of Agentic Superintelligence

Authors: Honglin Guo, Tao Gui, Yicheng Chen, Guanting Dong, Qiming Ge, Yuyang Hu, Zixian Huang, Jiajie Jin, Alexander Lam, Yining Li, Jiahang Lin, Yanjiang Liu, Xinyu Lu, Haijun Lv, Junlin Shang, Qisheng Su, Guoqiang Wang, Rui Wang, Zhecan Wang, Hao Xiang, Xinchen Xie, Shuhao Xing, Xiaoyu Xing, Wanghan Xu, Xinyu Yang, Yajie Yang, Chengfeng Zhao, Haoran Zhao, Ruojun Zhou, Yunhua Zhou, Yicheng Zou, Kun Cai, Qiye Cai, Xinmeng Che, Haodong Chen, Jiabei Chen, Jiahao Chen, Jiayi Chen, Yujia Chen, Lizhi Cui, Youheng Dai, Xin Deng, Yi Dong, Shihan Dou, Chenya Gu, Xu Guo, Ding Han, Feiyang Hao, Haotan He, Jie Hou, Binze Hu, Zijian Hu, Junhao Huang, Huicheng Jiang, Jiazhen Jiang, Shufan Jiang, Jiahao Kuang, Bowen Lai, Bo Li, Jiaqiang Li, Peng Li, Qilong Li, Zhuoqun Li, Jiaxiang Liu, Shuainan Liu, Tong Liu, Yi Liu, Zhonghang Lu, Jianwen Luo, Yanyi Luo, Huijie Lv, Ningsheng Ma, Zerun Ma, Houcheng Min, Chengjun Pan, Qiyuan Peng, Xiaoxuan Peng, Jianmin Qian, Jiantao Qiu, Wanying Ren, Huayu Sha, Jifei Shan, Zixin Shang, Bing Shao, Zhuohui Sheng, Jiayang Shi, Yang Shu, Aierpanjiang Simayi, Sirui Song, Yuxiao Song, Zhe Sun, Zhichao Sun, Wenzhe Tan, Wenhui Tian, Zhongbo Tian, Hanchen Wang, Pengbo Wang, Rui Wang, Yiding Wang, Yuhui Wang, Zhiheng Xi, Caijun Xu, Chao Xu, Yongfeng Xu, Xiaolei Yang, Zhixiong Yang, Qian Yao, Shihong Yi, Yuankai Ying, Jia Yu, Dingbo Yuan, Hao Yuan, Junjie Yuan, Bo Zhang, Caixian Zhang, Qiuyinzhe Zhang, Jiyuan Zhao, Penghao Zhao, Ying Zhao, Pujun Zheng, Xiaoxue Zhong, Xiaohao Zhou, Xinyu Zhou, Dongsheng Zhu, Guanru Zhu, Yulun Zhu, Yaojie Lu, Tao Ji, Hongyu Lin, Yutao Zhu, Pengfei Cao, Guoxiu He, Xianpei Han, Ben He, Zhicheng Dou, Kang Liu, Qi Zhang, Le Sun, Jun Zhao, Ji-Rong Wen, Xuanjing Huang, Yu-Gang Jiang, Bowen Zhou arXiv: arxiv.org/abs/2609.15818 Summary: As AI agents become participants in the development of their successors, they reshape both the production of intelligence and the role of human researchers. We introduce Atria Dawn Preview, a foundation agentic language model designed for scientific research and engineering workflows, with the goal of expanding the frontier of agent productivity in the real world. Trending because: 364 HuggingFace upvotes + trains a foundation agent through verifiable tool-mediated experience and examines human-AI research collaboration

2. ZGCM-1: A Fully Open and Extremely Efficient Foundation Model for Math and Agentic Search

Authors: Jiyan He, Guang Liang, Hao Liu, Haoxiang Guan, Jinbo Sun, Junyi Guo, Wenjun Feng, Yantai Xie, Yifei Shen, Bin Shao, Chuyang Wei, Kai Chen, Kexin Zhou, Minghang Zhu, Shuxin Zheng, Tie-Yan Liu, Taine Zhao, Wenhui Zhu, Xueyin Xu, Xiaoqing Zhang, Yatao Li, Yuxuan Ren arXiv: arxiv.org/abs/2609.13356 Summary: In this work, we present ZGCM-1, a fully open 7B dense foundation model trained from scratch with extreme data, system, and algorithmic efficiency. ZGCM-1 is founded on a core premise: compact models cannot passively memorize the open web, but can overcome parametric capacity limits by coupling deliberate internal thinking with active external tool use. Trending because: 289 HuggingFace upvotes + offers a fully open, efficient foundation model for mathematical reasoning and agentic search

3. Dream-RSI: Recursive Self-Improvement through Evolving Worlds

Authors: Tong Zheng, Xidong Wu, Zheng Zhang, Zhankui He, Chaoyi Zhang, Benjamin Coleman, Ruoqiao Wei, Di Bai, Haolin Liu, Rui Liu, Xue Wang, Yue Zhuan, Wang-Cheng Kang, Renkai Xiang, Heng Huang, Xinwu Cheng, Yunsong Guo arXiv: arxiv.org/abs/2609.14858 Summary: Recursive self-improvement is becoming increasingly vital for autonomous AI agents, where progress hinges on discovering high-value solutions across complex domains. The driver of this process is effective exploration, however, managing and improving exploration strategies remains a major bottleneck. Trending because: 253 HuggingFace upvotes + pursues recursive self-improvement by jointly evolving agents and the worlds that train them

4. PhysBrain 1.5: From Vision-Language Models to Physical Foundation Models

Authors: DeepCybo Team, Yu Bin, Haipeng Cao, Zheng Chang, Kai Chen, Youning Chen, Kailin Deng, Yichao Du, Xiaotong Fu, Haoyang Ge, Yunlong Guo, Chenliu Hao, Jiyan He, Xuguo He, Yakun Hou, Kai Hu, Cong Huang, Tuopusen Huang, Yu Huang, Hong Li, Peize Li, Shijie Lian, Xiaopeng Lin, Yun Lin, Haibao Liu, Haochen Liu, Qiuzhi Liu, Shengcai Liu, Zhiqiang Liu, Tao Luo, Peng Ren, Shuo Ren, Chaoyi Ruan, Zhaolong Shen, Yukun Shi, Qiyuan Su, Yuxuan Tian, Yining Wang, Changti Wu, Hao Wu, Xueyin Xu, Ruoqi Yang, Zhaoyang Yang, Hang Yuan, Zhaoyang Zeng, Hanwen Zhang, Ruimeng Zhang, Yao Zhang, Yibo Zhang, Yuxiang Zhang, Zhirui Zhang, Ziyi Zhang, Zubin Zheng, Zishen Zhuang arXiv: arxiv.org/abs/2609.14973 Summary: We present PhysBrain 1.5, a unified model for understanding physical environments, generating actions, and predicting future states. Motivated by the physical loop of observation, interaction, and environmental change, we bring these capabilities into a common learning framework. Trending because: 168 HuggingFace upvotes + extends vision-language modeling toward broadly capable physical foundation models

5. Grouped Value Attention: Efficient KV Caching via On-Demand Key Reconstruction

Authors: Vishesh Tripathi, Abhay Kumar, Ramsha Khan arXiv: arxiv.org/abs/2609.13285 Summary: The KV cache is a primary bottleneck for Transformer decoding: its memory footprint and cache-read traffic grow with sequence length. Grouped-query attention (GQA) reduces this cost by sharing key-value heads, but still stores both a key and a value at every step. Trending because: 57 HuggingFace upvotes + cuts KV-cache costs by reconstructing grouped keys on demand

6. LynnReal-Omni: Native multi-modal Video Generation for Agentic Visual Workflows

Authors: Xiaofeng Mao, Peijia Lin, Shaohao Rui, Yibo Zhang, Haibin Wan, Weijie Ma arXiv: arxiv.org/abs/2609.15863 Summary: Video diffusion models are stochastic and hard to control: precise content often requires repeated sampling without guaranteed success, and long-horizon scenes drift in appearance, interactions, and temporal coherence. Agentic visual creation provides explicit references, editable 3D scenes, or executable game states for stable control, but does not by itself guarantee high object or character fidelity. Trending because: 43 HuggingFace upvotes + unifies multimodal video generation for agentic visual workflows

7. RSIAgent: Autonomous Exploration for Recursive Self-improvement in New Environments

Authors: Sibo Zhu, Shicheng Fan, Xinyue Wang, Wenyi Wu, Kun Zhou, Biwei Huang arXiv: arxiv.org/abs/2609.15364 Summary: Digital agents must often adapt to new environments whose interfaces, tools, and failure modes are not fully captured by pretrained models. We introduce RSIAgent, a training-free multi-agent framework for recursive self-improvement through autonomous memory construction. Trending because: 23 HuggingFace upvotes + uses autonomous environment exploration to drive recursive self-improvement

8. BVB: Benchmarking Agentic Video Understanding via Programmatic Reconstruction in Blender

Authors: Yolo Y. Tang, Daiki Shimada, Jiayue Meng, Jing Bi, Pinxin Liu, Yicheng Wang, Yunzhong Xiao, Zhangyun Tan, Zeliang Zhang, Chao Huang, Susan Liang, Qianxiang Shen, Luchuan Song, Ali Vosoughi, Mingqian Feng, Melika Filvantorkaman, Chenliang Xu arXiv: arxiv.org/abs/2609.15478 Summary: Multimodal agents can create complex videos in software such as Blender by coding without relying on diffusion models. Yet video understanding benchmarks still evaluate models mainly through question answering. Trending because: 22 HuggingFace upvotes + benchmarks agentic video understanding through programmatic reconstruction in Blender

9. Discovery Foundation Models: Toward Open-Ended Discovery Intelligence

Authors: Ling Yang, Zhenfei Yin, Yingcheng Wu arXiv: arxiv.org/abs/2609.15973 Summary: Foundation models have progressed from learning and reasoning over existing knowledge, to increasingly learning through action, tool use, and outcome feedback. We argue that the next frontier is a further transition: from solving and acting within problems specified by humans to participating in the process by which new problems, representations, explanations, and knowledge are created. Trending because: 22 HuggingFace upvotes + frames foundation models as engines for open-ended scientific discovery

10. How Lossless Is Lossless Speculative Decoding? The Role of Numerical Precision in Orthrus

Authors: Ilya Koziev, Leonid Sinev, Ivan Oseledets arXiv: arxiv.org/abs/2609.15504 Summary: Orthrus is a hybrid autoregressive-diffusion architecture that accelerates autoregressive language-model inference by generating multiple tokens in parallel while using a frozen autoregressive backbone. Its central claim is that an intra-model consensus mechanism enables lossless speculative decoding, producing the same output sequence as the autoregressive model. Trending because: 21 HuggingFace upvotes + reveals numerical-precision caveats in supposedly lossless speculative decoding

11. AlayaVista: Streaming World Modeling from Panoramic States to Perspective Video

Authors: Jiaming Tan, Mingliang Zhai, Zhen Li, Yuwei Wu, Chuanhao Li, Kaipeng Zhang arXiv: arxiv.org/abs/2609.14462 Summary: Interactive video world models must maintain broad scene context under camera motion while producing high-fidelity observations with low latency. Existing approaches face a representation trade-off: perspective models operate on local views and must preserve off-screen content over long rollouts, whereas broader spatial coverage is typically obtained by synthesizing full-sphere videos or constructing explicit 3D representations. Trending because: 20 HuggingFace upvotes + streams world models from panoramic state representations to perspective video

12. Omni-Streaming Thinking

Authors: Enjun Du, Siyi Liu, Ziyu Zheng, Jingyu Li, Yiwen Guo, Yongqi Zhang, Difan Zou arXiv: arxiv.org/abs/2609.15128 Summary: Streaming omni-modal models must decide what and when to answer from the video chunks and synchronized audio observed so far. Visual cues often support an interpretation before an utterance or sound event is complete. Trending because: 16 HuggingFace upvotes + brings continuous test-time thinking to streaming multimodal interaction

13. Kaininja: Extending Native 3D Generators to the Part Level

Authors: Ruihan Yu, Lian Fu, Muyao Niu, Zheng-hui Huang, Yu-Ju Tsai, Sho Kuno, Fengbo Lan, Yonghao Yu, Erwin Wu, Ming-Hsuan Yang, Kaipeng Zhang, Zhixiang Wang arXiv: arxiv.org/abs/2609.15659 Summary: Native 3D generators turn one image into a single mesh. TRELLIS.2 and its peers deliver high-fidelity non-watertight geometry with materials, but the output is one fused object, while downstream work such as editing, rigging and simulation operates on part-level assets. Trending because: 16 HuggingFace upvotes + adds part-level generation and control to native 3D models

14. LLaDA-UI: Bringing Block-wise Diffusion to Vision-Language GUI Agents

Authors: Zhangxuan Gu, Haoxing Chen, Qi Qin, Yi Xin, Kai Gan, Lin Liu, Long Cui, Xiaomei Wang, Beitong Zhou, Yunzhu Zhang, Zhengwen Zeng, Changlong Gao, Weizhi Chen, Rongchao Zhang, Haoyuan Wu, Shuheng Shen, Changhua Meng, Weiqiang Wang, Jianguo Li, Zhenzhong Lan arXiv: arxiv.org/abs/2609.13287 Summary: Diffusion large language models (dLLMs) achieve high decoding efficiency through block-parallel, arbitrary-order generation, making them attractive for latency-sensitive applications. GUI agents represent a natural testbed for this paradigm, as they must repeatedly perceive screen states and emit structured, spatially grounded actions in real time. Trending because: 13 HuggingFace upvotes + applies block-wise diffusion modeling to vision-language GUI agents

15. HazardAuditor: From Executable Threats to Safer Computer-Use Agents

Authors: Yunhao Feng, Ruixiao Lin, Ming Wen, Yanming Guo, Xingjun Ma, Yutao Wu, Xinhao Deng, Shouling Ji arXiv: arxiv.org/abs/2609.15134 Summary: Computer-use agents increasingly interact with browsers, terminals, file systems, and external services, introducing safety risks that emerge through runtime behavior rather than generated content alone. Existing guard models target static prompts and responses and are poorly suited to agent execution; existing executable safety platforms produce evaluation verdicts rather than the normalized supervision a guard model needs to learn across heterogeneous agent frameworks. Trending because: 13 HuggingFace upvotes + turns executable computer-use threats into practical agent safety audits

16. Agent as Policy for Robotic Manipulation

Authors: Mengzhao Jia, Yang Lin, Xixin Zhang, Zhihan Zhang, Xiaobai Liu, Meng Jiang arXiv: arxiv.org/abs/2609.12541 Summary: We demonstrate that a general-purpose agent can directly drive a physical robot throughout task execution without any task-specific or environment-specific training. We introduce Agent as Policy (AGP), which places task planning and execution under the agent’s control. Trending because: 11 HuggingFace upvotes + treats an agent itself as the policy for robotic manipulation

17. When Agents Slow Down: Understanding LLM Agents’ Test-Time Strategies via Elo-per-token Analysis

Authors: Kaiyuan Liu, Qiuyang Mang, Bo Peng, Wenhao Chai, Hanchen Li, Shreyas Pimpalgaonkar, Luke Zettlemoyer, Alex Dimakis, Alvin Cheung arXiv: arxiv.org/abs/2609.15309 Summary: Large language model (LLM) agents allocate test-time compute adaptively as they revise solutions, use tools, explore alternatives, and decide when to stop. This test-time strategy makes it difficult to measure how agent performance scales. Trending because: 11 HuggingFace upvotes + measures the efficiency of agent test-time strategies with Elo per token

18. Not All Prompts Are Equal: Exploration-Guided Prompt Scaffolding for Multimodal Reinforcement Post-Training

Authors: Yuanhao Yue, Qianli Ma, Chengyu Wang, Haoting Wang, Lei Shen, Jun Huang arXiv: arxiv.org/abs/2609.15051 Summary: Training prompts in online reinforcement learning (RL) differ substantially in how informative they are for the current policy: some are already saturated while others are too difficult to yield reliable learning signals, yet both receive equal rollout budget under standard training. We propose an exploration-guided prompt scaffolding framework that adapts the training prompt distribution dynamically throughout RL post-training of multimodal large language models (MLLMs). Trending because: 6 HuggingFace upvotes + optimizes multimodal reinforcement-learning prompts through guided exploration

19. MInTRL: Off-policy Intervention can boost On-policy RL

Authors: Mingyu Chen, Yefan Tao, Gerald Friedland, Xuezhou Zhang, Chris Kong arXiv: arxiv.org/abs/2609.12419 Summary: Reinforcement learning with verifiable rewards is typically performed on-policy, keeping training data close to the current policy but limiting learning to trajectories that the policy can discover itself. Off-policy methods such as supervised fine-tuning, on the other hand, can leverage external knowledge beyond the base model’s capabilities, but may suffer from large distribution shift. Trending because: 5 HuggingFace upvotes + shows how off-policy intervention data can strengthen on-policy reinforcement learning

20. Attention-DP3: Spatially Object-aware 3D Diffusion Policy via Geometry-aligned Attentional Conditioning

Authors: Changbo Yan, Zhongbo Zhang, Zaibin Zhang, Yifan Wang, Lijun Wang, Huchuan Lu arXiv: arxiv.org/abs/2609.13318 Summary: 3D point-cloud observations are inherently ambiguous in complex, cluttered manipulation scenes, where target objects may be partially occluded or tightly intermingled with visually similar distractors. As a result, standard 3D diffusion policies often struggle to localize and exploit task-relevant geometry as scene complexity grows. Trending because: 5 HuggingFace upvotes + adds geometry-aligned object attention to 3D diffusion policies