Monthly collection

November 2025

The papers that drew the most attention during this month of AI research.

Papers: 73
Citations: 4.8K
Research labs: 6

Explore the year

View all months →

Aug2025 Sep2025 Oct2025 Nov2025 Dec2025 Jan2026 Feb2026 Mar2026 Apr2026 May2026 Jun2026 Jul2026 Aug2026

1,016papers
curated and indexed

73 papers published in November 2025

01

arXiv.org

From Code Foundation Models to Agents and Applications: A Comprehensive Survey and Practical Guide to Code Intelligence

This survey comprehensively examines the lifecycle of code-focused large language models (LLMs), from data curation and pre-training to post-training, alignment, and deployment as autonomous agents. It analyzes both general-purpose LLMs (e.g., GPT-4, Claude, LLaMA) and code-specialized models (e.g., StarCoder, Code LLaMA, DeepSeek-Coder, QwenCoder),…

Jian Yang, Xianglong Liu, Weifeng Lv, Ken Deng, et al.

Published: Nov 2025
Upvotes: 306
Citations: 12

Read summary Original

02

arXiv.org

Z-Image: An Efficient Image Generation Foundation Model with Single-Stream Diffusion Transformer

Z-Image is a 6B-parameter image generation foundation model from Alibaba Group, built on a Scalable Single-Stream Diffusion Transformer (S3-DiT). It challenges the 'scale-at-all-costs' paradigm by optimizing data infrastructure, architecture, training, and inference. The full training workflow costs 314K H800 GPU hours (~$628K). Z-Image-Turbo, a distilled…

Z-Image Team, Huanqia Cai, Sihan Cao, Ruoyi Du, et al.

Published: Nov 2025
Upvotes: 249
Citations: 206

Read summary Original

03

arXiv.org

Thinking with Video: Video Generation as a Promising Multimodal Reasoning Paradigm

The paper proposes 'Thinking with Video', a new paradigm using video generation models like Sora-2 for multimodal reasoning, addressing limitations of text- and image-based paradigms. The authors introduce VideoThinkBench, a benchmark with vision-centric tasks (eyeballing puzzles, visual puzzles, ARC-AGI-2, mazes) and text-centric tasks (GSM8K, MATH, MMLU,…

Jingqi Tong, Yurong Mou, Hangcheng Li, Mingzhe Li, et al.

Published: Nov 2025
Upvotes: 242
Citations: 27

Read summary Original

04

arXiv.org

Kandinsky 5.0: A Family of Foundation Models for Image and Video Generation

Kandinsky 5.0 is a family of foundation models for high-resolution image and video synthesis, comprising three model line-ups: Image Lite (6B parameters), Video Lite (2B), and Video Pro (19B). The models are built on a unified latent diffusion architecture with a CrossDiT backbone and trained using flow matching. Key innovations include the NABLA sparse…

Vladimir Arkhipkin, Vladimir Korviakov, Nikolai Gerasimenko, Denis Parkhomenko, et al.

Published: Nov 2025
Upvotes: 235
Citations: 9

Read summary Original

05

arXiv.org

Lumine: An Open Recipe for Building Generalist Agents in 3D Open Worlds

Lumine, developed by ByteDance Seed, is the first open recipe for building generalist agents that complete hours-long missions in real time within 3D open-world environments. Built on Qwen2-VL-7B-Base, it processes raw pixels at 5 Hz and generates keyboard-mouse actions at 30 Hz using action chunking, with a hybrid thinking mode that invokes reasoning only…

Weihao Tan, Xiangyang Li, Yunhao Fang, Heyuan Yao, et al.

Published: Nov 2025
Upvotes: 219
Citations: 13

Read summary Original

06

arXiv.org

MiroThinker: Pushing the Performance Boundaries of Open-Source Research Agents via Model, Context, and Interactive Scaling

MiroThinker v1.0 is an open-source research agent that introduces interaction scaling as a third performance dimension, alongside model size and context length. It trains models via reinforcement learning to handle deeper and more frequent agent–environment interactions, using a 256K context window and up to 600 tool calls per task. The 72B variant…

MiroMind Team, Song Bai, Lidong Bing, Carson Chen, et al.

Published: Nov 2025
Upvotes: 197
Citations: 62

Read summary Original

07

arXiv.org

LongVT: Incentivizing "Thinking with Long Videos" via Native Tool Calling

LongVT is an end-to-end agentic framework that enables large multimodal models (LMMs) to reason over long videos by interleaving multimodal Chain-of-Tool-Thought (iMCoTT) with native video cropping tool calls. It mimics human global-to-local viewing: the model first skims the video, then invokes a crop_video tool to inspect specific temporal windows, and…

Zuhao Yang, Sudong Wang, Kaichen Zhang, Keming Wu, et al.

Published: Nov 2025
Upvotes: 188
Citations: 51

Read summary Original

08

arXiv.org

General Agentic Memory Via Deep Research

The paper introduces General Agentic Memory (GAM), a memory framework for AI agents that follows a just-in-time (JIT) compilation principle, contrasting with traditional ahead-of-time (AOT) static memory systems that suffer from information loss. GAM uses a dual-agent design: a Memorizer that compresses key historical information into lightweight memory…

B. Y. Yan, Chaofan Li, Hongjin Qian, Shuqi Lu, et al.

Published: Nov 2025
Upvotes: 173
Citations: 34

Read summary Original

09

Qwen

Qwen3-VL Technical Report

Qwen3-VL is a state-of-the-art vision-language model family from the Qwen team, released on December 1, 2025. It supports interleaved contexts up to 256K tokens and comes in dense (2B/4B/8B/32B) and MoE (30B-A3B/235B-A22B) variants. Key architectural innovations include interleaved-MRoPE for balanced spatial-temporal encoding, DeepStack for multi-level ViT…

Shuai Bai, Yuxuan Cai, Ruizhe Chen, Keqin Chen, et al.

Published: Nov 2025
Upvotes: 164
Citations: 1.8K

Read summary Original

10

arXiv.org

Tiny Model, Big Logic: Diversity-Driven Optimization Elicits Large-Model Reasoning Ability in VibeThinker-1.5B

This technical report from Sina Weibo Inc. introduces VibeThinker-1.5B, a 1.5B-parameter dense model that challenges the assumption that small models lack robust reasoning capabilities. Developed with a post-training methodology based on the 'Spectrum-to-Signal Principle (SSP)', the approach decouples SFT and RL objectives: the SFT 'Spectrum Phase' uses…

Sen Xu, Yi Zhou, Wei Wang, Jixin Min, et al.

Published: Nov 2025
Upvotes: 142
Citations: 8

Read summary Original

11

Meta AI

Souper-Model: How Simple Arithmetic Unlocks State-of-the-Art LLM Performance

The paper introduces Soup Of Category Experts (SoCE), a model souping technique that improves LLM performance by leveraging benchmark composition and non-uniform weighted averaging. SoCE identifies weakly-correlated benchmark categories, selects expert models for each, and optimizes weights to combine them. Experiments on BFCL, MGSM, and ∞-Bench show SoCE…

Shalini Maiti, Amar Budhiraja, Bhavul Gauri, Gaurav Chaurasia, et al.

Published: Nov 2025
Upvotes: 140
Citations: 4

Read summary Original

12

Meta AI

SAM 3: Segment Anything with Concepts

SAM 3 is a unified model for promptable concept segmentation (PCS) in images and videos, accepting noun phrases, image exemplars, or both as prompts to detect, segment, and track all matching instances. It decouples recognition and localization via a presence head, improving detection accuracy. A data engine with human and AI verifiers produced 4M unique…

Nicolas Carion, Laura Gustafson, Yuan-Ting Hu, Shoubhik Debnath, et al.

Published: Nov 2025
Upvotes: 138
Citations: 711

Read summary Original

13

arXiv.org

P1: Mastering Physics Olympiads with Reinforcement Learning

The P1 technical report introduces a family of open-source physics reasoning models trained entirely via reinforcement learning (RL). P1-235B-A22B is the first open-source model to achieve gold-medal performance at the International Physics Olympiad 2025 (IPhO 2025), scoring 21.2/30 and ranking 3rd behind Gemini-2.5-Pro and GPT-5. It won 12 gold and 1…

Jiacheng Chen, Qianjia Cheng, Fangchen Yu, Haiyuan Wan, et al.

Published: Nov 2025
Upvotes: 135
Citations: 4

Read summary Original

14

arXiv.org

Diffusion Language Models are Super Data Learners

The paper demonstrates that under data-constrained pre-training, diffusion language models (DLMs) consistently surpass autoregressive (AR) models of equal size, a phenomenon termed the 'Intelligence Crossover.' This crossover shifts later with more or higher-quality data, earlier with larger models, and persists across dense and sparse architectures. The…

Jinjie Ni, Qian Liu, Longxu Dou, Chao Du, et al.

Published: Nov 2025
Upvotes: 132
Citations: 50

Read summary Original

15

arXiv.org

One Small Step in Latent, One Giant Leap for Pixels: Fast Latent Upscale Adapter for Your Diffusion Models

The paper introduces the Latent Upscaler Adapter (LUA), a lightweight module that performs super-resolution directly on a diffusion model's latent code before VAE decoding, enabling high-resolution image synthesis without retraining the generator or adding diffusion stages. LUA uses a shared SwinIR-style backbone with scale-specific pixel-shuffle heads for…

Aleksandr Razin, Danil Kazantsev, Ilya Makarov

Published: Nov 2025
Upvotes: 132
Citations: 1

Read summary Original

16

NVIDIA

TiDAR: Think in Diffusion, Talk in Autoregression

TiDAR is a sequence-level hybrid architecture that combines diffusion and autoregressive (AR) language modeling in a single forward pass using structured attention masks. It drafts tokens in parallel via diffusion (thinking) and samples final outputs autoregressively (talking), exploiting free GPU compute slots to achieve high throughput without quality…

Jingyu Liu, Xin Dong, Zhifan Ye, Rishabh Mehta, et al.

Published: Nov 2025
Upvotes: 130
Citations: 33

Read summary Original

17

arXiv.org

Latent Collaboration in Multi-Agent Systems

LatentMAS is a training-free framework enabling multi-agent systems (MAS) to collaborate entirely in the continuous latent space, bypassing text-based communication. Each agent generates latent thoughts via auto-regressive last-layer hidden states, and a shared latent working memory (KV caches) transfers these representations losslessly to subsequent…

Jiaru Zou, Ruizhong Qiu, Gaotang Li, Xiyuan Yang, et al.

Published: Nov 2025
Upvotes: 130
Citations: 35

Read summary Original

18

NVIDIA

ToolOrchestra: Elevating Intelligence via Efficient Model and Tool Orchestration

The paper introduces ToolOrchestra, a method for training small language models as orchestrators that coordinate diverse tools, including basic tools (web search, code interpreter), specialized LLMs (math, coding), and generalist LLMs (GPT-5, Claude Opus 4.1). The resulting 8B-parameter Orchestrator model is trained end-to-end with reinforcement learning…

Hongjin Su, Shizhe Diao, Ximing Lu, Mingjie Liu, et al.

Published: Nov 2025
Upvotes: 129
Citations: 32

Read summary Original

19

arXiv.org

GigaEvo: An Open Source Optimization Framework Powered By LLMs And Evolution Algorithms

GigaEvo is an open-source framework for LLM-driven evolutionary computation, inspired by AlphaEvolve, designed to enhance reproducibility and experimentation. It features modular components: a Redis-based storage for evolutionary units, an asyncio DAG execution engine, a MAP-Elites quality-diversity algorithm with single- and multi-island support, and a…

Valentin Khrulkov, Andrey Galichin, Denis Bashkirov, Dmitry Vinichenko, et al.

Published: Nov 2025
Upvotes: 122
Citations: 13

Read summary Original

20

Meta AI

SAM 3D: 3Dfy Anything in Images

SAM 3D is a generative model for 3D object reconstruction from a single image, predicting geometry, texture, and layout. It excels in natural images with occlusion and clutter, using a human- and model-in-the-loop pipeline to create large-scale 3D annotation data. The model uses a multi-stage training framework: synthetic pretraining on 2.7M meshes…

SAM 3D Team, Xingyu Chen, Fu-Jen Chu, Pierre Gleize, et al.

Published: Nov 2025
Upvotes: 117
Citations: 197

Read summary Original

21

arXiv.org

VIDEOP2R: Video Understanding from Perception to Reasoning

The paper introduces VIDEOP2R, a process-aware reinforcement fine-tuning (RFT) framework for video understanding that explicitly separates perception and reasoning. It addresses limitations of existing video RFT methods, which treat video reasoning as a single process and assign a single final reward, leading to poor credit assignment. VIDEOP2R consists of…

Yifan Jiang, Yueying Wang, Rui Zhao, Toufiq Parag, et al.

Published: Nov 2025
Upvotes: 113
Citations: 8

Read summary Original

22

arXiv.org

Think-at-Hard: Selective Latent Iterations to Improve Reasoning Language Models

The paper introduces Think-at-Hard (TaH), a looped transformer that selectively applies latent iterations to improve reasoning in small language models. The authors identify a 'latent overthinking' problem where always iterating can flip correct predictions into errors. An oracle policy that iterates only on mispredicted tokens boosts accuracy by up to…

Tianyu Fu, Yichen You, Zekai Chen, Guohao Dai, et al.

Published: Nov 2025
Upvotes: 110
Citations: 18

Read summary Original

23

arXiv.org

Agent0: Unleashing Self-Evolving Agents from Zero Data via Tool-Integrated Reasoning

Agent0 is a fully autonomous framework that evolves LLM agents from zero external data by combining tool-integrated reasoning with multi-step co-evolution. It initializes two agents from the same base LLM: a curriculum agent trained via GRPO to generate frontier tasks, rewarded by the executor's uncertainty (self-consistency), tool-use frequency, and a…

Peng Xia, Kaide Zeng, Jiaqi Liu, Can Qin, et al.

Published: Nov 2025
Upvotes: 110
Citations: 54

Read summary Original

24

arXiv.org

Grounding Computer Use Agents on Human Demonstrations

The paper introduces GROUNDCUA, a large-scale desktop grounding dataset built from expert human demonstrations, covering 87 applications across 12 categories with 56K screenshots and over 3.56M human-verified element annotations. The dataset features high-resolution images (0.39–7.0 megapixels), dense annotations (average 64 per screenshot), and small…

Aarash Feizi, Shravan Nayak, Xiangru Jian, Kevin Qinghong Lin, et al.

Published: Nov 2025
Upvotes: 107
Citations: 9

Read summary Original

25

arXiv.org

Uni-MoE-2.0-Omni: Scaling Language-Centric Omnimodal Large Model with Advanced MoE, Training and Data

Uni-MoE-2.0-Omni is a fully open-source omnimodal large model (OLM) built from the dense Qwen2.5-7B LLM, designed for unified understanding, reasoning, and generation across text, image, audio, and video. Its architecture introduces a dynamic-capacity Mixture-of-Experts (MoE) with shared, routed, and null experts for efficient computation and modality…

Yunxin Li, Xinyu Chen, Shenyuan Jiang, Haoyuan Shi, et al.

Published: Nov 2025
Upvotes: 107
Citations: 21

Read summary Original

26

arXiv.org

VCode: a Multimodal Coding Benchmark with SVG as Symbolic Visual Representation

The paper introduces VCode, a benchmark that reframes multimodal understanding as visual-centric code generation: given an image, a model must produce SVG code that preserves symbolic meaning for downstream reasoning. VCode covers three domains: general commonsense (MM-Vet), professional disciplines (MMMU), and visual-centric perception (CV-Bench),…

Kevin Qinghong Lin, Yuhao Zheng, Hangyu Ran, Dantong Zhu, et al.

Published: Nov 2025
Upvotes: 104
Citations: 12

Read summary Original

27

arXiv.org

Depth Anything 3: Recovering the Visual Space from Any Views

Depth Anything 3 (DA3) is a model that predicts spatially consistent geometry from any number of images, with or without known camera poses. It uses a single plain transformer (e.g., vanilla DINOv2) as backbone, with an input-adaptive cross-view self-attention mechanism and a dual-DPT head that jointly outputs depth and ray maps. A depth-ray representation…

Haotong Lin, Sili Chen, Junhao Liew, Donny Y. Chen, et al.

Published: Nov 2025
Upvotes: 103
Citations: 499

Read summary Original

28

arXiv.org

V-Thinker: Interactive Thinking with Images

V-Thinker is a general-purpose multimodal reasoning assistant that enables interactive, vision-centric thinking through end-to-end reinforcement learning. It comprises a Data Evolution Flywheel that automatically synthesizes, evolves, and verifies interactive reasoning datasets across diversity, quality, and difficulty, producing the V-Interaction-400K…

Runqi Qiao, Qiuna Tan, Minghan Yang, Guanting Dong, et al.

Published: Nov 2025
Upvotes: 98
Citations: 19

Read summary Original

29

arXiv.org

DoPE: Denoising Rotary Position Embedding

The paper introduces DOPE (Denoising Rotary Position Embedding), a training-free method to improve length extrapolation in LLMs by mitigating instabilities caused by RoPE. The authors show via spectral analysis that RoPE's low-frequency components cause spectral amplification, leading to low-rank, over-aligned attention patterns and attention sinks. DOPE…

Jing Xiong, Liyang Fan, Hui Shen, Zunhai Su, et al.

Published: Nov 2025
Upvotes: 98
Citations: 2

Read summary Original

30

arXiv.org

GeoVista: Web-Augmented Agentic Visual Reasoning for Geolocalization

The paper introduces GeoVista, an agentic multimodal model for geolocalization that integrates image zoom-in and web-search tools within a dynamic reasoning loop. The authors curate GeoBench, a benchmark of 1,142 high-resolution images (photos, panoramas, satellite) from 66 countries, with multi-level labels (country, province, city) and haversine distance…

Yikun Wang, Zuyan Liu, Ziyi Wang, Han Hu, et al.

Published: Nov 2025
Upvotes: 98
Citations: 8

Read summary Original

31

arXiv.org

OpenMMReasoner: Pushing the Frontiers for Multimodal Reasoning with an Open and General Recipe

OpenMMReasoner introduces a fully transparent two-stage recipe for training large multimodal reasoning models (LMRMs), covering supervised fine-tuning (SFT) and reinforcement learning (RL). The SFT stage uses an 874K-sample cold-start dataset built from 103K raw questions, distilled with a strong teacher model (Qwen3-VL-235B-Instruct) and scaled via ×8…

Kaichen Zhang, Keming Wu, Zuhao Yang, Bo Li, et al.

Published: Nov 2025
Upvotes: 96
Citations: 31

Read summary Original

32

DeepSeek

DeepSeekMath-V2: Towards Self-Verifiable Mathematical Reasoning

DeepSeekMath-V2 is a large language model for natural-language theorem proving, built on DeepSeek-V3.2-Exp-Base, that achieves self-verifiable mathematical reasoning. The authors argue that final-answer rewards are insufficient because correct answers do not guarantee correct reasoning and are inapplicable to theorem proving. They train a verifier using…

Zhihong Shao, Yuxiang Luo, Chengda Lu, Z. Z. Ren, et al.

Published: Nov 2025
Upvotes: 96
Citations: 60

Read summary Original

33

arXiv.org

HaluMem: Evaluating Hallucinations in Memory Systems of Agents

HaluMem is the first operation-level benchmark for evaluating hallucinations in memory systems for LLMs and AI agents. It defines three tasks—memory extraction, memory updating, and memory question answering—to localize hallucinations to specific operational stages. The benchmark includes two datasets, HaluMem-Medium and HaluMem-Long, with about 15k memory…

Ding Chen, Simin Niu, Kehang Li, Peng Liu, et al.

Published: Nov 2025
Upvotes: 95
Citations: 37

Read summary Original

34

arXiv.org

AutoEnv: Automated Environments for Measuring Cross-Environment Agent Learning

The paper introduces AUTOENV, an automated framework for generating heterogeneous agent environments by treating them as factorizable distributions over transitions, observations, and rewards, with three abstraction layers (BaseEnv, ObsEnv, SkinEnv) and a three-stage verification pipeline. Using AUTOENV, the authors construct AUTOENV-36, a dataset of 36…

Jiayi Zhang, Yiran Peng, Fanqi Kong, Cheng Yang, et al.

Published: Nov 2025
Upvotes: 92
Citations: 13

Read summary Original

35

Conference of the European Chapter of the Association for Computational Linguistics

Unveiling Intrinsic Dimension of Texts: from Academic Abstract to Creative Story

This paper provides the first comprehensive study of intrinsic dimension (ID) in text representations, grounding it in interpretable properties through cross-encoder analysis, linguistic features, and sparse autoencoders (SAEs). The authors establish three key findings: (1) ID is complementary to entropy-based metrics, as after controlling for length, the…

Vladislav Pedashenko, Laida Kushnareva, Yana Khassan Nibal, Eduard Tulchinskii, et al.

Published: Nov 2025
Upvotes: 91
Citations: 1

Read summary Original

36

arXiv.org

Scaling Agent Learning via Experience Synthesis

DreamGym is a unified framework for scaling reinforcement learning (RL) for LLM-based agents by synthesizing diverse experiences, addressing the high costs and limitations of real-environment rollouts. It uses a reasoning-based experience model that operates in an abstract textual state space, generating consistent state transitions and reward signals via…

Zhaorun Chen, Zhuokai Zhao, Kai Zhang, Bo Liu, et al.

Published: Nov 2025
Upvotes: 83
Citations: 24

Read summary Original

37

arXiv.org

PAN: A World Model for General, Interactable, and Long-Horizon World Simulation

PAN is a general, interactable, and long-horizon world model that predicts future world states via video simulation conditioned on history and natural language actions. It uses the Generative Latent Prediction (GLP) architecture, combining an autoregressive LLM-based latent dynamics backbone (Qwen2.5-VL-7B) with a video diffusion decoder (Wan2.1-T2V-14B)…

PAN Team, Jiannan Xiang, Yi Gu, Zihan Liu, et al.

Published: Nov 2025
Upvotes: 82
Citations: 33

Read summary Original

38

Research paper

IterResearch: Rethinking Long-Horizon Agents with Interaction Scaling

IterResearch is a novel iterative deep-research paradigm that addresses the limitations of mono-contextual approaches, which accumulate all information in a single expanding context window, leading to context suffocation and noise contamination. It uses an MDP-inspired architecture with strategic workspace reconstruction, maintaining an evolving report as…

Guoxin Chen, Zile Qiao, Xuanzhong Chen, Donglei Yu, et al.

Published: Nov 2025
Upvotes: 80
Citations: 14

Read summary Original

39

Conference of the European Chapter of the Association for Computational Linguistics

Multimodal Evaluation of Russian-language Architectures

The paper introduces MERA Multi, the first open multimodal evaluation benchmark for Russian-language architectures, addressing the lack of such benchmarks for Slavic languages. It comprises 18 instruction-based tasks across text, image, audio, and video modalities, built on a unified taxonomy of multimodal abilities. The benchmark includes 11 private…

Artem Chervyakov, Ulyana Isaeva, Anton Emelyanov, Artem Safin, et al.

Published: Nov 2025
Upvotes: 79
Citations: 3

Read summary Original

40

arXiv.org

Reasoning via Video: The First Evaluation of Video Models' Reasoning Abilities through Maze-Solving Tasks

This paper introduces VR-Bench, a benchmark for evaluating the reasoning abilities of video generation models through maze-solving tasks. It comprises 7,920 procedurally generated videos across five maze types (Regular, Irregular, 3D, Trapfield, Sokoban) with varying difficulty and textures. The authors propose a 'reasoning via video' paradigm, where…

Cheng Yang, Haiyuan Wan, Yiran Peng, Xin Cheng, et al.

Published: Nov 2025
Upvotes: 78
Citations: 15

Read summary Original

41

arXiv.org

AraLingBench A Human-Annotated Benchmark for Evaluating Arabic Linguistic Capabilities of Large Language Models

AraLingBench is a fully human-annotated benchmark for evaluating the Arabic linguistic competence of large language models (LLMs). It consists of 150 expert-designed multiple-choice questions across five categories: grammar, morphology, spelling, reading comprehension, and syntax. The benchmark was constructed by five Arabic linguistics experts from the…

Mohammad Zbeeb, Hasan Abed Al Kader Hammoud, Sina Mukalled, Nadine Rizk, et al.

Published: Nov 2025
Upvotes: 74
Citations: 1

Read summary Original

42

arXiv.org

Deep Research: A Systematic Survey

This survey provides a comprehensive overview of Deep Research (DR) systems, which combine large language models with external tools to perform complex, open-ended research tasks. The authors formalize a three-stage roadmap for DR: Agentic Search, Integrated Research, and Full-stack AI Scientist. They identify four key components of DR systems: query…

Zhengliang Shi, Yiqun Chen, Haitao Li, Weiwei Sun, et al.

Published: Nov 2025
Upvotes: 73
Citations: 15

Read summary Original

43

arXiv.org

MMaDA-Parallel: Multimodal Large Diffusion Language Models for Thinking-Aware Editing and Generation

The paper introduces MMaDA-Parallel, a parallel multimodal diffusion framework for thinking-aware image editing and generation. The authors identify a failure mode in sequential autoregressive approaches where pre-generation reasoning can degrade performance due to error propagation. They propose ParaBench, a benchmark evaluating both text and image…

Ye Tian, Ling Yang, Jiongfan Yang, Anran Wang, et al.

Published: Nov 2025
Upvotes: 72
Citations: 8

Read summary Original

44

arXiv.org

Part-X-MLLM: Part-aware 3D Multimodal Large Language Model

Part-X-MLLM is a native 3D multimodal large language model that unifies diverse 3D tasks by formulating them as programs in a structured, executable grammar. Given an RGB point cloud and a natural language prompt, the model autoregressively generates a token sequence encoding part-level bounding boxes, semantic descriptions, and edit commands. This…

Chunshi Wang, Junliang Ye, Yunhan Yang, Yang Li, et al.

Published: Nov 2025
Upvotes: 72
Citations: 5

Read summary Original

45

arXiv (Cornell University)

Back to Basics: Let Denoising Generative Models Denoise

The paper argues that denoising diffusion models should directly predict clean images (x-prediction) rather than noise (epsilon-prediction) or velocity (v-prediction), as natural data lies on a low-dimensional manifold while noised quantities do not. The authors propose 'Just image Transformers' (JiT), a plain Vision Transformer applied to large pixel…

Tianhong Li, Kaiming He

Published: Nov 2025
Upvotes: 71
Citations: 4

Read summary Original

46

arXiv.org

DeCo: Frequency-Decoupled Pixel Diffusion for End-to-End Image Generation

DeCo is a frequency-decoupled pixel diffusion framework for end-to-end image generation. It addresses the challenge of pixel diffusion models jointly modeling high-frequency signals and low-frequency semantics in a single diffusion transformer (DiT), which slows training and inference. DeCo uses a DiT to model low-frequency semantics from downsampled…

Zehong Ma, Longhui Wei, Shuai Wang, Shiliang Zhang, et al.

Published: Nov 2025
Upvotes: 66
Citations: 37

Read summary Original

47

arXiv.org

LFM2 Technical Report

LFM2 is a family of Liquid Foundation Models designed for efficient on-device deployment, using a hardware-in-the-loop architecture search to create a hybrid backbone of gated short convolutions and a few grouped-query attention blocks. The family includes dense models (350M–2.6B) and an 8.3B MoE variant (1.5B active), all with 32K context. Pre-trained on…

Alexander Amini, Anna Banaszak, Harold Benoit, Arthur Böök, et al.

Published: Nov 2025
Upvotes: 65
Citations: 23

Read summary Original

48

arXiv.org

DR Tulu: Reinforcement Learning with Evolving Rubrics for Deep Research

The paper introduces DR Tulu-8B, the first fully open model trained end-to-end for long-form deep research tasks, using a new method called Reinforcement Learning with Evolving Rubrics (RLER). RLER constructs and maintains rubrics that co-evolve with the policy model during training, incorporating newly explored information from search and contrasting…

Rulin Shao, Akari Asai, Shannon Zejiang Shen, Hamish Ivison, et al.

Published: Nov 2025
Upvotes: 64
Citations: 75

Read summary Original

49

arXiv.org

A Style is Worth One Code: Unlocking Code-to-Style Image Generation with Discrete Style Space

The paper introduces CoTyle, the first open-source framework for code-to-style image generation, where a numerical style code controls the visual style of generated images. Existing methods rely on textual prompts, reference images, or LoRA fine-tuning, which suffer from poor consistency, limited creativity, or complex representations. CoTyle trains a…

Huijie Liu, Shuhao Cui, Haoxiang Cao, Shuai Ma, et al.

Published: Nov 2025
Upvotes: 63
Citations: 0

Read summary Original

50

arXiv.org

When Visualizing is the First Step to Reasoning: MIRA, a Benchmark for Visual Chain-of-Thought

The paper introduces MIRA (Multimodal Imagination for Reasoning Assessment), a benchmark designed to evaluate multimodal large language models (MLLMs) on tasks that require generating or using intermediate visual images (e.g., sketches, diagrams) for successful reasoning, mirroring human 'drawing to think'. MIRA contains 546 problems across 20 task types…

Yiyang Zhou, Haoqin Tu, Zijun Wang, Zeyu Wang, et al.

Published: Nov 2025
Upvotes: 60
Citations: 10

Read summary Original

51

arXiv.org

Step-Audio-R1 Technical Report

Step-Audio-R1 is the first audio reasoning model to successfully benefit from deliberate thinking, addressing the historical issue where audio language models performed worse with extended reasoning. The team identifies the root cause as 'textual surrogate reasoning,' where models reason from transcripts rather than acoustic features. They propose…

Fei Tian, Xiangyu Tony Zhang, Yuxin Zhang, Haoyang Zhang, et al.

Published: Nov 2025
Upvotes: 60
Citations: 35

Read summary Original

52

Annual Meeting of the Association for Computational Linguistics

GroupRank: A Groupwise Paradigm for Effective and Efficient Passage Reranking with LLMs

GroupRank is a novel groupwise reranking paradigm for LLM-based passage retrieval that balances the efficiency of pointwise methods with the accuracy of listwise approaches. It partitions candidate documents into small groups for parallel cross-document comparison, avoiding context-window limits and reducing latency. Training uses an answer-free data…

Meixiu Long, Duolin Sun, Dan Yang, Yihan Jiao, et al.

Published: Nov 2025
Upvotes: 59
Citations: 3

Read summary Original

53

Meta AI

What Does It Take to Be a Good AI Research Agent? Studying the Role of Ideation Diversity

This paper investigates the role of ideation diversity in AI research agents' performance on MLE-bench, a benchmark of 75 Kaggle machine learning tasks. The authors analyzed 11,000 agent trajectories from 6 LLM backbones and 3 scaffolds, measuring ideation diversity via Shannon entropy of model architectures in initial drafts. They found that…

Alexis Audran-Reiss, Jordi Armengol-Estapé, Karen Hambardzumyan, Amar Budhiraja, et al.

Published: Nov 2025
Upvotes: 59
Citations: 5

Read summary Original

54

arXiv.org

Time-to-Move: Training-Free Motion Controlled Video Generation via Dual-Clock Denoising

Time-to-Move (TTM) is a training-free, plug-and-play framework for motion- and appearance-controlled video generation using image-to-video (I2V) diffusion models. It uses crude reference animations (e.g., cut-and-drag or depth-based reprojection) as motion cues, adapting SDEdit's noise injection to video. To preserve appearance, it anchors generation to…

Assaf Singer, Noam Rotstein, Amir Mann, Ron Kimmel, et al.

Published: Nov 2025
Upvotes: 58
Citations: 11

Read summary Original

55

Annual Meeting of the Association for Computational Linguistics

Too Good to be Bad: On the Failure of LLMs to Role-Play Villains

This paper investigates the ability of large language models (LLMs) to role-play morally ambiguous or villainous characters, hypothesizing that safety alignment conflicts with authentic antagonistic portrayal. The authors introduce the Moral RolePlay benchmark, a dataset with a four-level moral alignment scale (Moral Paragons, Flawed-but-Good, Egoists,…

Zihao Yi, Qingxuan Jiang, Ruotian Ma, Xingyu Chen, et al.

Published: Nov 2025
Upvotes: 57
Citations: 7

Read summary Original

56

Conference on Empirical Methods in Natural Language Processing

MADD: Multi-Agent Drug Discovery Orchestra

The paper introduces MADD, a multi-agent system for automated hit identification in early drug discovery. MADD uses four specialized agents (Decomposer, Orchestrator, Summarizer, Chat Agent) to process natural language queries, decompose them into subtasks, generate molecules using GAN and transformer-based CVAE models, and predict properties like IC50 and…

Gleb V. Solovev, Alina B. Zhidkovskaya, Anastasia Orlova, Nina Gubina, et al.

Published: Nov 2025
Upvotes: 57
Citations: 4

Read summary Original

57

arXiv.org

V-ReasonBench: Toward Unified Reasoning Benchmark Suite for Video Generation Models

V-ReasonBench is a benchmark for evaluating reasoning in generative video models under the Chain-of-Frame paradigm, where the final frame represents the model's answer. It covers four reasoning dimensions: structured problem-solving (arithmetic, code execution, Sudoku, Tic-Tac-Toe), spatial cognition (shape fitting, visual symmetry, color connection),…

Yang Luo, Xuanlei Zhao, Baijiong Lin, Lingting Zhu, et al.

Published: Nov 2025
Upvotes: 56
Citations: 17

Read summary Original

58

arXiv.org

MedSAM3: Delving into Segment Anything with Medical Concepts

MedSAM-3 adapts the SAM 3 architecture for medical image and video segmentation, enabling Promptable Concept Segmentation (PCS) via open-vocabulary text prompts. The model is fine-tuned on medical images paired with concise concept phrases (≤3 words) across modalities like X-ray, MRI, Ultrasound, CT, and video. The MedSAM-3 Agent integrates a Multimodal…

Anglin Liu, Rundong Xue, Xu R. Cao, Yifan Shen, et al.

Published: Nov 2025
Upvotes: 56
Citations: 28

Read summary Original

59

arXiv.org

UniAVGen: Unified Audio and Video Generation with Asymmetric Cross-Modal Interactions

UniAVGen is a unified framework for human-centric joint audio and video generation, addressing limitations in existing methods like poor lip synchronization and semantic inconsistency. It uses a dual-branch architecture with two parallel Diffusion Transformers (DiTs) for video and audio, enabling a cohesive cross-modal latent space. The core innovation is…

Guozhen Zhang, Zixiang Zhou, Teng Hu, Ziqiao Peng, et al.

Published: Nov 2025
Upvotes: 54
Citations: 28

Read summary Original

60

arXiv.org

Black-Box On-Policy Distillation of Large Language Models

The paper introduces Generative Adversarial Distillation (GAD), a method for black-box distillation of large language models (LLMs) where only the teacher's text outputs are accessible. GAD frames the student as a generator and trains a discriminator to distinguish student from teacher responses, forming a minimax game. The discriminator acts as an…

Tianzhu Ye, Li Dong, Zewen Chi, Xun Wu, et al.

Published: Nov 2025
Upvotes: 54
Citations: 43

Read summary Original

61

arXiv.org

Computer-Use Agents as Judges for Generative User Interface

The paper introduces AUI-Gym, a benchmark for automatic GUI development and testing, and a Coder-CUA collaboration framework. AUI-Gym includes 52 applications across six domains (app, landing, game, interactive, tool, utility) with 1,560 tasks synthesized by GPT-5 and human-validated. Each task is paired with a rule-based verifier that programmatically…

Kevin Qinghong Lin, Siyuan Hu, Linjie Li, Zhengyuan Yang, et al.

Published: Nov 2025
Upvotes: 54
Citations: 6

Read summary Original

62

arXiv.org

First Frame Is the Place to Go for Video Content Customization

This paper introduces FFGo, a lightweight add-on for video generation models that enables multi-reference video content customization without architectural changes or large-scale fine-tuning. The authors discover that pre-trained video models treat the first frame as a conceptual memory buffer, storing visual entities for later reuse. FFGo leverages this…

Jingxi Chen, Zongxia Li, Zhichao Liu, Guangyao Shi, et al.

Published: Nov 2025
Upvotes: 54
Citations: 8

Read summary Original

63

arXiv.org

What about gravity in video generation? Post-Training Newton's Laws with Verifiable Rewards

The paper introduces NewtonRewards, a physics-grounded post-training framework for video generation that enforces Newton's laws of motion using verifiable rewards. It addresses the issue that video diffusion models often produce visually realistic but physically implausible motion. The method uses optical flow as a proxy for velocity and visual features…

Minh-Quan Le, Yuanzhi Zhu, Vicky Kalogeiton, Dimitris Samaras

Published: Nov 2025
Upvotes: 54
Citations: 18

Read summary Original

64

arXiv.org

Visual Spatial Tuning

The paper introduces Visual Spatial Tuning (VST), a framework to enhance the spatial perception and reasoning abilities of Vision-Language Models (VLMs) without adding specialized 3D encoders. VST comprises two datasets: VST-P, with 4.1 million samples across 19 tasks covering single-image, multi-image, and video scenarios, and VST-R, with 135K samples for…

Rui Yang, Ziyu Zhu, Yanwei Li, Jingjia Huang, et al.

Published: Nov 2025
Upvotes: 53
Citations: 64

Read summary Original

65

arXiv.org

DRIVE: Data Curation Best Practices for Reinforcement Learning with Verifiable Reward in Competitive Code Generation

This paper from Tencent's Hunyuan Team presents a two-stage reinforcement learning with verifiable reward (RLVR) framework for competitive programming code generation, addressing the underexplored area of data curation and curriculum design. The method begins with SFT distilled from strong open-source models, then applies two RL stages: first, entropy…

Speed Zhu, Jianwei Cai, Guang Chen, Lulu Wu, et al.

Published: Nov 2025
Upvotes: 53
Citations: 2

Read summary Original

66

arXiv.org

PhysX-Anything: Simulation-Ready Physical 3D Assets from Single Image

PhysX-Anything is a novel framework that generates simulation-ready physical 3D assets from a single in-the-wild image. It uses a VLM-based generative model to predict geometry, articulation, and physical properties, and introduces a voxel-based representation that reduces token count by 193x, enabling explicit geometry learning without special tokens. The…

Ziang Cao, Fangzhou Hong, Zhaoxi Chen, Liang Pan, et al.

Published: Nov 2025
Upvotes: 53
Citations: 25

Read summary Original

67

OpenAI

Weight-sparse transformers have interpretable circuits

The paper introduces weight-sparse transformers, where most weights are zero, to improve mechanistic interpretability. By constraining the L0 norm, models learn disentangled circuits for tasks, which are pruned to isolate minimal circuits. These circuits are compact and contain neurons and residual channels corresponding to natural concepts, with…

Leo Gao, Achyuta Rajaram, Jacob Coxon, Soham V. Govande, et al.

Published: Nov 2025
Upvotes: 0
Citations: 37

Read summary Original

68

Together AI

Escaping the Verifier: Learning to Reason via Demonstrations

The paper introduces RARO (Relativistic Adversarial Reasoning Optimization), a method to train large language models (LLMs) to reason using only expert demonstrations, without task-specific verifiers or human preferences. RARO frames the problem as inverse reinforcement learning, setting up an adversarial game between a policy and a relativistic critic…

Locke Cai, Ivan Provilkov

Published: Nov 2025
Upvotes: Not indexed
Citations: 5

Read summary Original

69

Together AI

Opportunistic Expert Activation: Batch-Aware Expert Routing for Faster Decode Without Retraining

The paper introduces Opportunistic Expert Activation (OEA), a batch-aware routing framework that reduces Mixture-of-Experts (MoE) decode latency without retraining. During autoregressive generation, MoE models become memory-bound at moderate batch sizes because latency scales with the number of unique activated experts. OEA operates in two phases: first,…

Costin-Andrei Oncescu, Qingyang Wu, Wai Tong Chung, Robert Wu, et al.

Published: Nov 2025
Upvotes: Not indexed
Citations: 4

Read summary Original

70

Together AI

CDLM: Consistency Diffusion Language Models for Faster Sampling

CDLM (Consistency Diffusion Language Models) is a training-based method to accelerate inference in Diffusion Language Models (DLMs) by addressing two bottlenecks: excessive refinement steps and incompatibility with KV caching. It integrates consistency modeling to reduce sampling steps via multi-token finalization and enforces a block-wise causal attention…

Minseo Kim, Chenfeng Xu, Coleman Hooper, Harman M. Singh, et al.

Published: Nov 2025
Upvotes: Not indexed
Citations: 17

Read summary Original

71

Together AI

Beat the Long Tail: Distribution-Aware Speculative Decoding for RL Training

Reinforcement learning (RL) post-training for large language models is bottlenecked by the rollout phase, which accounts for over 70% of training time. The authors identify a long-tail distribution of rollout lengths, where a few long generations dominate wall-clock time, and note that historical rollouts reveal stable prompt-level patterns across epochs.…

Zelei Shao, Vikranth Srivatsa, Sanjana Srivastava, Qingyang Wu, et al.

Published: Nov 2025
Upvotes: Not indexed
Citations: 9

Read summary Original

72

Together AI

Kitty: Accurate and Efficient 2-bit KV Cache Quantization with Dynamic Channel-Wise Precision Boost

The paper introduces Kitty, an algorithm-system co-design for 2-bit KV cache quantization that maintains accuracy close to FP16 while reducing memory by nearly 8x. The authors observe that 4-bit quantization preserves accuracy, but 2-bit degrades it significantly. They propose Dynamic Channel-wise Precision Boost, which identifies critical key-cache…

Haojun Xia, Xiaoxia Wu, Jisen Li, Robert Wu, et al.

Published: Nov 2025
Upvotes: Not indexed
Citations: 7

Read summary Original

73

Together AI

ParallelKittens: Systematic and Practical Simplification of Multi-GPU AI Kernels

ParallelKittens (PK) is a minimal CUDA framework that simplifies the development of overlapped multi-GPU kernels by distilling three key principles: transfer mechanisms, scheduling strategies, and design overheads. It extends ThunderKittens with eight core primitives and a unified programming template, enabling efficient compute-communication overlap. PK…

Stuart H. Sul, Simran Arora, B. Spector, Christopher R'e

Published: Nov 2025
Upvotes: Not indexed
Citations: 9

Read summary Original