Today's research,
properly prioritized.

Spatial intelligence is becoming a foundation for embodied agents, robotic planning, and multimodal assistants. To improve the spatial reasoning ability of VLM agents, existing work has mainly followed two lines. One line uses post-training methods, such as supervised fine-tuning and reinforcement learning. Another line adopts an agentic paradigm in which the model calls external spatial tools, such as depth estimation and 3D reconstruction tools, to gather intermediate spatial evidence. We study a complementary and underexplored route: Can a frozen VLM agent improve its spatial reasoning through parameter-update-free self-evolution, without depending on external expert spatial tools at inference time? We present Spatial Memory Agent (SMA), an experience-grounded runtime framework that converts verified spatial experience into reusable transferable lessons. In a verifiable spatial environment, SMA queries the frozen VLM, obtains a predicted answer and reward, and uses verifier-guided reflection to distill compact transferable lessons from spatial experience. SMA further assigns each lesson a Transfer Reliability Score (TRS), which is initialized uniformly and calibrated from later retrieval outcomes as visit evidence of future transfer reliability. During read-only deployment, SMA retrieves lessons by semantic filter and similarity-TRS combined ranking, allowing the retrieved memory to guide frozen model inference. Across five representative spatial benchmarks and four base VLMs, SMA achieves the highest macro average in every base-model block and the best accuracy among the evaluated methods in most of the 20 evaluations, establishing a practical parameter-update-free path for spatial self-evolution across the evaluated frozen model scales and environments.

Previsualization is an intermediate layer between ideas and production in film, games, architecture, and urban design. It lets creators iteratively refine scenes, actions, cameras, and spatial-temporal dynamics. Yet existing generative methods rely on simple prompts to jointly control all of these factors through one-shot image or video synthesis, offering weak controllability and limited support for iterative editing. Fundamentally, a world comprises multiple elements with geometry, appearance, and other attributes, together with cameras. Different frames are produced through local modifications or recombinations of this shared state, which is otherwise largely reused. Therefore, we argue that the missing component is an explicit and persistent working state. To address this, we present StateFlow, a state-centric framework for generative previsualization. Rather than generating videos in one shot, StateFlow uses an editable 3D world to organize scene structure, evolution, and cameras, while off-the-shelf video models enhance visual quality when higher fidelity is desired. This world is maintained as a persistent structured 3D state of scene elements and camera configurations, serving as the core working representation for previsualization. Built on this insight, StateFlow has three stages to construct, evolve, and access the world state. State construction lifts generated 2D content into a coherent 3D world through prior-guided, conflict-aware dual-view initialization, while State evolution translates user intent into structured state transitions while preserving world memory, avoiding full-scene regeneration for each edit. State access uses render-feedback reflection to refine camera plans into visually feasible trajectories, avoiding reliance on VLM semantics alone. Experiments show that StateFlow produces high-quality 3D worlds for video creation and game-like prototyping.

Artist-grounded image generation requires more than appending an artist name to a prompt. Image models often respond to artist names through canonical shortcuts, such as recurring motifs, generic palettes, or overrepresented period signatures, rather than preserving the user's intended scene. We introduce Atelier, a shortcut-aware control-state planning framework for artist-grounded image generation. Atelier translates underspecified artistic intent into an explicit control state that separates scene anchors, preserve/transform decisions, style-regime hypotheses, role-bound artist evidence, and shortcut-avoidance constraints. It grounds this state using artist-level knowledge and local patch references, compiles backend-aware generation plans, and iteratively refines candidates through global and local authenticity feedback. We further introduce ArtIntentBench, a benchmark covering Van Gogh and Qi Baishi across artwork re-rendering, period/style-controlled generation, historically unseen subjects, shortcut auditing, and human preference evaluation. Across open-weight and closed-source generators, Atelier improves artist-level style fidelity, preserves source structure more faithfully, and substantially reduces shortcut substitution compared with prompt-engineered, retrieval-augmented, and general-purpose agent baselines. These results suggest that artist-grounded generation is bottlenecked not only by image synthesis, but by the upstream inference of explicit, evidence-grounded artistic controls.

Pose-driven human animation synthesizes a video of a target person from a single reference image and a driving pose stream. Real-time generation is essential for interactive applications such as live streaming, telepresence, and virtual avatars, yet diffusion-based systems require minutes to hours per clip, precluding responsive interaction. We present LiveAnimate, to our knowledge the first animation system to combine real-time streaming with stable long-form generation at billion scale, built on a 14B-parameter video Diffusion Transformer (DiT). A two-stage training pipeline first adapts a pretrained bidirectional DiT into a block-causal autoregressive generator through Reference-Anchored Teacher-Forcing Adaptation, and then reduces the sampling budget to three steps through Block-wise Self-Forcing Distillation. To preserve appearance over extended streams, we introduce Pose-Retrieval Sink Attention (PR-Sink), a bounded KV-cache mechanism combining a Static Sink that permanently anchors the first generated block, a Dynamic Sink that holds a pose-retrieved historical block, and a three-slot Rolling Window. When a pose recurs, PR-Sink restores the relevant appearance context without retaining the entire sequence, so memory and per-block latency remain constant regardless of stream duration. Together with Ulysses sequence parallelism and operator fusion, these designs enable 19.63\,FPS streaming inference on two NVIDIA H100 GPUs. On a three-minute benchmark, LiveAnimate maintains nearly constant perceptual quality and identity from the first 30 seconds to the final minute, while prior systems degrade substantially or require hours of offline computation for the same rollout. These results establish a new operating point in quality, latency, and duration for interactive full-body animation.

Interactive world models must support persistent memory, responsive interaction, and long-horizon generation, yet these requirements place conflicting demands on the model. Maintaining history in the denoiser context or key-value cache incurs growing cost, forcing a trade-off between session length and retained memory, while low-latency interaction relies on few-step generation whose capabilities are bounded by its teacher. Evoke addresses both limitations by externalizing persistent world state and redesigning the teacher for long-horizon interactive generation. Scene geometry is maintained in an external, camera-indexed world state bank, from which only view-relevant information is retrieved, keeping the denoiser context bounded as the session grows. Rather than treating the teacher as a fixed generator, we design it for long-horizon supervision: its sparse attention combines chunk-wise grouping, retrieval of selected distant frames, and a linear-attention global state, yielding linear growth in memory and compute while enabling supervision over long horizons. Such supervision exposes content drift that stays locally plausible within short windows, while per-chunk conditioning enables prompt changes and event control throughout the sequence. A 30-second distribution-matching objective, applied under self-forced rollouts, transfers both capabilities to a three-step student that uses no classifier-free guidance, improving resistance to long-term drift while preserving responsive conditioning. With bounded context and recurrent external memory, Evoke supports open-ended, continuously evolving generation; on a single H200 at 384times 640, each 1.5,s chunk is generated in 2.11,s. As a three-step world model, Evoke achieves state-of-the-art performance on WBench while remaining competitive on VBench-Long and VBench-2.0.

We introduce , a recurrent Transformer architecture with fixed-size memory that generalizes sliding-window attention while remaining parallelizable during training. consists of two coupled models: a prefiller Q, which leverages full attentionIn practice, we use interleaved full and sliding-window attention for Q, as this yields stronger performance. The essential requirement is that Q be more expressive than P, with access to the full history. to produce memory targets m'_t, and a decoder P, which uses only sliding-window attention and recurrent K/V injection to produce decoder memories m_t for next-token prediction. We train with a memory consistency loss that aligns m_t with m'_t, allowing inference to use P alone. Empirically, improves validation loss and downstream pretraining benchmarks over sliding-window and latent recurrent transformer baselines. Moreover, sharing parameters between P and Q reduces parameter memory while preserving most of the gains.

Recent work on distillation transfers the capabilities of large models to smaller ones often by updating the latter's parameters, through teacher forcing, on-policy distillation, and related training-time methods. In this paper, we ask whether such transfer can instead occur at test time. We study strong-to-weak scaffolding: whether a stronger builder model can construct inference-time harnesses that help a weaker target model solve tasks more reliably without any parameter updates. Using four representative Theory-of-Mind benchmarks, each builder model uses 5% of the data as a validation set to iteratively refine its harness over multiple rounds, after which the finalized harness is evaluated on the full test set. Empirically, this form of test-time capability transfer is highly effective, nearly doubling average target-model performance from 0.49 to 0.91. Our analysis shows that the gains come primarily from offloading unstable model reasoning into deterministic code, benchmark-specific routing, and strict answer-format enforcement, rather than from encouraging the target model to reason more extensively or sample more broadly. We further find that builder-model reasoning effort improves harness quality monotonically, platform effects are modest relative to the builder model's own capability, and weaker target models receive the largest gains. These results suggest that inference-time harness design is an important complement to conventional training-time distillation, enabling strong models to transfer cognitive structure to weaker models without retraining.

Agent skills represent a standardized format for packaging procedural knowledge and domain expertise, serving within agent harness systems as an essential mechanism to continually constrain a language model's behavior space for repeatable, high-quality task execution. However, because strong closed-source models entail high inference costs, current popular agent harnesses, such as Codex and OpenClaw, remain prohibitively expensive when deploying these skills to accomplish real-world tasks. The rapid capability enhancement of open-source models deployable on consumer-grade GPUs presents a compelling opportunity to drastically reduce these costs by leveraging skill-based behavioral constraints. Nevertheless, automatically generating effective skills tailored specifically for such compact models remains a significant practical challenge. To address this, we propose SKILLER, a natural-language-driven reinforcement learning framework designed to automatically generate executor-specific skills for small models, which employs a strong model as the actor and critic, treats the small-model agent system as the environment, and propagates all reinforcement learning signals entirely via natural language. Extensive experimental evaluations across five relevant benchmarks using Qwen3.5-9B and Qwen3.5-4B demonstrate that SKILLER outperforms three open-source and one closed-source skill generation or evolution methods, achieving absolute gains ranging from 4.3 to 20.4 percentage points for the 9B model and 1.8 to 13.3 points for the 4B model, while remarkably matching the performance of strong closed-source models on single-skill tasks in SkillsBench. The project is available at https://github.com/DANG-ai/SKILLER.

Large Language Models (LLMs) increasingly act as agents whose procedural knowledge is stored in reusable skill packages and loaded at inference time. As skill libraries grow, a central challenge is to expose the smallest sufficient executable context under a limited context budget. Existing systems struggle to reuse routines below the whole-skill level, preserve procedural contracts during compression, keep compressed routines executable and expandable, and update the compressed library as skills evolve. These challenges reveal a unit mismatch: skills are retrieved as packages, compressed as text, and converted into execution graphs only after retrieval, whereas reliable reuse requires a contract-bearing procedural unit. We propose SkillZip, an execution-aware procedural abstraction framework that performs contract-preserving compression over section-level graphs. SkillZip rewrites recurring contract-valid motifs into reversible ported macros while preserving boundary signatures, dependency closure, verifier reachability, and source-level expansion. At inference time, it hydrates a compact, dependency-closed context and expands macros only when required. ReZip further integrates new skills and revises risky macros using execution evidence. Comprehensive experiments1 on technical and embodied agent benchmarks show SkillZip consistently outperforms the strongest baseline by up to 12.2 points, while achieving a 3.46x compression ratio with 99.2% dependency preservation and 98.7% verifier reachability. Scaling analyses further confirm robust retrieval across skill libraries ranging from 200 to 100K skills.

Scientific discovery increasingly requires AI systems that can reason over scientific evidence of heterogeneous modalities, interact with scientific tools and environments, and sustain progress across long task horizons. We present Intern-S2-Preview, a series of scientific agentic foundation models designed to support multimodal scientific understanding, reasoning, generation, and long-horizon tasks. The training pipeline begins with scientific multimodal pre-training over rendered scientific documents, interleaved image-text data, and diverse scientific corpora. Starting from the pretrained checkpoint, we apply a unified post-training pipeline consisting of supervised fine-tuning, scalable multi-task reinforcement learning (RL), black- and white-box agentic RL, and on-policy distillation. This pipeline is supported by practical techniques that improve rollout and training stability and efficiency, including partial rollout with off-policy correction, adaptive length regularization, online speculative decoding, robust multi-task optimization, and trace-aware experience assembly for agentic tasks. At the architecture level, Intern-S2-Preview-397B extends time series modelling from efficient long-sequence understanding to numerical forecasting, while Memory Decoder is studied as a separate memory-augmented path for rapid scientific specialization without modifying the frozen 397B backbone. Evaluations across scientific, multimodal, agentic, and general-purpose benchmarks show that Intern-S2-Preview-397B achieves competitive or leading results in multiple settings. The time series modules improve scientific signal understanding and forecasting on SciTS, while the separate Intern-MemDec-4B extension improves the Biology-Instructions average score from 56.92 to 60.32 without modifying the frozen 397B backbone.

Autoregressive transformers compute along two axes: horizontally across generated tokens, and vertically through model depth. Dense attention gives each token broad horizontal access to the past, but the vertical feedback channel between decoding steps remains narrow: only the sampled token returns to the bottom of the stack, while the top-layer hidden state is discarded. We introduce the full-bandwidth transformer, which widens this channel with latent feedback: at each decoding step, the previous top-layer hidden state is fused with the sampled token embedding through a gated linear unit and fed back as the next input. Latent feedback lets non-verbalized computation re-enter the stack with a renewed depth budget, while preserving the standard transformer architecture, KV cache, and language-modeling objective. To train full-bandwidth transformers without losing parallel teacher forcing, we use a scheduled multi-pass objective that introduces latent feedback late in pretraining and mixes a small fraction of deeper feedback passes for stability. We train 1B-parameter full-bandwidth transformers up to 400B tokens and find that latent feedback improves validation loss, 5-shot language-model evaluation, math and coding generation, and instruction-tuned performance. With negligible per-token decoding overhead, full-bandwidth transformers match or approach standard transformers trained with roughly 1.5times more tokens, and manage to produce shorter reasoning traces at equal or better accuracy.

No single large language model (LLM) is optimal across all queries and budget constraints, making model routing essential for cost-effective deployment. Existing routers adopt diverse formulations and implementations, making fair comparison and extension difficult. We present a unified formulation of LLM routing as a sequential decision process characterized by five components: context encoders, model encoders, scoring functions, decision rules, and learning signals, covering single-turn, multi-turn, and personalized routing. Based on this formulation, we develop an automated pipeline for constructing routing supervision and evaluating routers jointly on response quality and inference cost. The resulting benchmark, xRouteBench, spans generic LLM, memory-augmented, vision, time-series, and personalized routing tasks. We further introduce LLMRouter, an open-source modular infrastructure with more than 16 representative routers. Our empirical study shows that learned routers outperform the strongest fixed-model baseline by 14.6% relatively, lightweight routers become more competitive under tight cost constraints, and user-conditioned routing consistently improves personalization.

Test-time compute scaling is a primary driver of performance in large reasoning models (LRMs), but extreme inefficiency bounds current approaches, shifting the critical question from how much compute to spend, to where to allocate it. We formalize test-time reasoning as a constrained compute allocation problem over partial trajectories. Under a fixed hardware budget, existing paradigms fail to actively allocate the compute to the most promising partial progress: traditional parallel sampling treats traces independently and induces severe memory bottlenecks, while subtractive pruning starves hardware and fails to actively and sufficiently shift the output distribution. To overcome this dichotomy, we introduce Gambit, an inference algorithm that executes thought-level beam search. By periodically pruning unpromising trajectories and immediately branching from high-quality prefixes, Gambit dynamically concentrates compute onto the most promising reasoning traces via a light-weight scorer probing hidden states while maintaining continuous high hardware utilization. Extensive evaluations across multiple models and benchmarks demonstrate that Gambit strictly dominates existing baselines. Under identical hardware constraints, our method yields up to a +6.7\% absolute accuracy gain on HMMT-24 and +3.3\% on AIME-25 over pruning baselines, delivers >2times higher throughput on trace completion, and reduces total token consumption by up to 68.5\% relative to standard parallel sampling.

Visual-token pruning can substantially reduce the inference cost of multimodal large language models (MLLMs), yet existing methods largely rely on fixed, handcrafted heuristics and costly expert trial and error. As pruning objectives, budgets, and model architectures diversify, manually navigating the expanding design space becomes increasingly difficult. This paper aims to build an AI4AI framework for visual-token pruning by addressing a natural question: Can large language models automatically design effective visual-token reduction algorithms? Although LLMs possess broad algorithmic knowledge and strong reasoning capabilities, translating such general knowledge into effective solutions for a specialized task remains nontrivial. We argue that the key lies in designing an appropriate search-state representation that connects the internal knowledge of LLMs with the structural requirements and constraints of visual-token pruning. Based on this insight, we propose AutoPrune, a training-free framework for LLM-driven visual-token pruning policy design. At its core, AutoPrune introduces a Token Pruning Domain-Specific Language (TPDSL) comprising 131 reusable atoms for budget control, token scoring, selection constraints, and token reassembly. A key property of TPDSL is that it represents each search state as a residual modification of a strong base policy. This residual formulation narrows the search space and directs the LLM's attention toward the policy components that are most consequential for performance. Experiments on 14 multimodal benchmarks and three MLLM backbones demonstrate the effectiveness, efficiency, and transferability of AutoPrune. Even when removing 94.4% of visual tokens, AutoPrune preserves more than 99% of full-token performance while reducing FLOPs by 9.9x and prefill latency by 6.4x.

本文介绍多模态商业构想基准，包含六个领域的三万条样本，每个领域都包含仅靠文字难以传达的视觉线索。作者通过低秩适配监督微调和组相对策略优化训练盲测与已知标准代理，并使用创造性、可行性及业务评价标准作为奖励；实验显示，两类代理都明显优于文字描述和多模态基线。

“用图像思考”让多模态大模型能够主动裁剪和缩放图像，但这些操作相较直接推理通常只带来很小甚至负面的收益，并增加令牌成本。作者将视觉工具调用形式化为因果图，并通过策略、轨迹和单步干预审计观察证据的作用；在六个模型和五个细粒度感知基准上，他们发现部分模型虽然总体准确率提升，却在广泛轨迹中没有因果有效地使用视觉证据，具体表现为调用却未查看，以及查看却未规划。

智能体能力不仅取决于模型权重，也取决于提示词、工具、技能和控制流组成的外壳；该工作将外壳自我进化建模为种群选择，并冻结模型权重，通过保留扩展契约、替代谱系档案以及统一的证据驱动编辑接口来避免回退。四个基准上的结果显示，一轮进化平均增加约十七个百分点，包括终端任务、网页真实任务和软件工程任务的提升，且一个终端任务外壳可原样迁移到经过验证的软件工程基准。

世界模型的架构、训练目标与状态表示相互影响，尚无适用于各种环境的统一方案，因此论文构建了一个面向自主研究编码智能体的闭环基准。该基准覆盖八个游戏环境，使用统一的结构化真实实体状态表示，让智能体在固定算力预算内改进给定世界模型；在六十四次会话中，科迪克斯五点四和克劳德奥普斯四点六有六十三次取得改进，百分之九十一的获胜修改是新的目标函数、状态表示、展开流程或架构变化，而非超参数调整。

该工作提出一种面向大语言模型智能体的长期记忆方法，以语义片段级整合替代逐轮整合：系统将多轮交互组成片段，并把每个完成的片段编码为上下文无关的类型化记忆记录，再通过结构化索引进行查询规划和证据检索。实验显示，该方法在两个长期记忆基准上取得百分之八十九点二二和百分之九十二点二零的成绩，相比一种现有方法分别减少百分之八十六点零和百分之七十五点九的记忆构建词元，同时不增加查询阶段的词元使用量。

空间智能对于具身智能体、机器人规划和多模态助手日益重要，但现有方法主要依赖监督微调、强化学习或推理时调用深度估计和三维重建等外部空间工具。本文提出空间记忆智能体，通过验证器引导反思将空间经验转化为可迁移记忆，并依据后续检索结果校准迁移可靠性分数，在冻结视觉语言模型、不更新参数且推理时不调用外部专家工具的条件下引导推理；在五个空间基准和四个基础模型上的实验显示，该方法在所有基础模型分组中取得最高宏平均，并在多数评测中达到最高准确率。

文章指出，数字智能体主要改变软件状态、具身智能体主要改变物理状态，而两者都没有把人的状态变化与自主权作为建模、干预和评估的核心。文章提出以多模态事件感知、纵向可纠正记忆、个人世界模型和遵守同意与安全约束的干预策略构成闭环，并进一步讨论评估指标、基准、边缘个人模型和治理要求。

Pose-driven human animation synthesizes a video of a target person from a single reference image and a driving pose stream. Real-time generation is essential for interactive applications such as live streaming, telepresence, and virtual avatars, yet diffusion-based systems require minutes to hours per clip, precluding responsive interaction. We present LiveAnimate, to our knowledge the first animation system to combine real-time streaming with stable long-form generation at billion scale, built on a 14B-parameter video Diffusion Transformer (DiT). A two-stage training pipeline first adapts a pretrained bidirectional DiT into a block-causal autoregressive generator through Reference-Anchored Teacher-Forcing Adaptation, and then reduces the sampling budget to three steps through Block-wise Self-Forcing Distillation. To preserve appearance over extended streams, we introduce Pose-Retrieval Sink Attention (PR-Sink), a bounded KV-cache mechanism combining a Static Sink that permanently anchors the first generated block, a Dynamic Sink that holds a pose-retrieved historical block, and a three-slot Rolling Window. When a pose recurs, PR-Sink restores the relevant appearance context without retaining the entire sequence, so memory and per-block latency remain constant regardless of stream duration. Together with Ulysses sequence parallelism and operator fusion, these designs enable 19.63\,FPS streaming inference on two NVIDIA H100 GPUs. On a three-minute benchmark, LiveAnimate maintains nearly constant perceptual quality and identity from the first 30 seconds to the final minute, while prior systems degrade substantially or require hours of offline computation for the same rollout. These results establish a new operating point in quality, latency, and duration for interactive full-body animation.

Artist-grounded image generation requires more than appending an artist name to a prompt. Image models often respond to artist names through canonical shortcuts, such as recurring motifs, generic palettes, or overrepresented period signatures, rather than preserving the user's intended scene. We introduce Atelier, a shortcut-aware control-state planning framework for artist-grounded image generation. Atelier translates underspecified artistic intent into an explicit control state that separates scene anchors, preserve/transform decisions, style-regime hypotheses, role-bound artist evidence, and shortcut-avoidance constraints. It grounds this state using artist-level knowledge and local patch references, compiles backend-aware generation plans, and iteratively refines candidates through global and local authenticity feedback. We further introduce ArtIntentBench, a benchmark covering Van Gogh and Qi Baishi across artwork re-rendering, period/style-controlled generation, historically unseen subjects, shortcut auditing, and human preference evaluation. Across open-weight and closed-source generators, Atelier improves artist-level style fidelity, preserves source structure more faithfully, and substantially reduces shortcut substitution compared with prompt-engineered, retrieval-augmented, and general-purpose agent baselines. These results suggest that artist-grounded generation is bottlenecked not only by image synthesis, but by the upstream inference of explicit, evidence-grounded artistic controls.

Transforming multimodal sources into condensed and structured media outputs can be fundamentally conceptualized as a long-horizon agentic process centered on a model-harness system. While an ideal harness system should align with human design priors and accumulate reusable experience through empirical exploration to drive recursive self-improvement, existing paradigms remain static and fall short of this capability. In this paper, we present AutoDesign, a framework that aligns with human design priors, where a meta-harness optimizer guides a code agent to recursively improve harness based on rollout feedback. To instantiate and evaluate this framework, we focus on the academic paper-to-poster generation task and introduce PosterBench, comprising a 100-paper Main Track spanning five disciplines and PosterBench-mini, a shared 10-paper subset for controlled evaluation. On the PosterBench Main Track, AutoDesign achieves the highest score of 78.32, surpassing the closed-source commercial system Claude Design by 7.45 points. Across seven controlled code-agent-model configurations, integrating the learned DesignHarness consistently improves performance, increasing the average PosterBench Score from 54.99 to 67.39 (+12.4%). In a fully autonomous long-horizon loop, it executes 253 tool calls and 11 editing turns within 40 minutes for under $3, reaching average conference-poster quality in human evaluation. A system-blind human study further demonstrates that AutoDesign achieves the highest human preference among evaluated systems.

An LLM agent's capability depends not only on model weights but on its harness: prompts, tools, skills, and control flow. Self-improvement loops already edit harnesses, yet single-lineage search is path-dependent and local wins often regress other tasks. We introduce DarwinX, which treats self-evolution as selection over a population of harnesses with the model frozen: a preserve-and-extend contract admits only variants that extend coverage without regressing, an archive keeps alternative lineages for recombination, and failure-, teacher-, and self-derived evidence share one edit interface. Fitness comes from each benchmark's own verifier: no gold solutions, no hand-picked winners. Across four benchmarks that progressively separate the evolution signal from the test, one loop adds about 17 points on average: Terminal-Bench 2.1 rises +7.7 to 83.2% on a matched base and to the verified frontier at 84.7% on a stronger one; TerminalWorld's held-out split reaches 68.3%, ahead of every off-the-shelf agent; WebArena-Infinity real-task pass@1 rises from 43.5% to 93.0% audit-clean; and a Terminal-Bench 2.1 harness transfers unchanged to SWE-bench Verified. What evolves is general agent competence, not benchmark-specific patches, so it survives changes of task, verifier, and base model. A frozen model need not be a fixed agent: harness selection turns evaluation compute into durable capability.

We present the first systematic study of Massive activations (MAs) in layer-interleaved HLA LLMs and uncover two architecture-aligned morphologies: MAs consistently spike immediately before full attention layers, forming pre-attention spikes (PAS), and can persist through intervening linear attention layers, giving rise to inter-spike plateaus (ISP). As full attention becomes denser, successive PAS become increasingly connected through ISP, ultimately recovering the stable MA morphology of full attention LLMs. We establish the recurrence of this organization across five linear attention architectures, six hybridization configurations, five data domains, and representative open-source hybrid models spanning 1.2B to 397B total parameters. Controlled pretraining of GDN-based hybrids at scales up to 1.3B shows that both morphologies emerge early and respond asymmetrically to output gating: full attention output gating strongly attenuates their absolute magnitudes without eliminating their layerwise organization, whereas removing GDN gates yields comparatively modest amplification. Mechanistically, our systematic-outlier analysis supports a shared lifecycle account governed by the timing of MA cancellation. PAS follows a localized write-sink-cancel process, while the extended persistence of ISP is consistent with delayed cancellation. At the full attention limit, this account recovers the stable MA morphology characteristic of full attention LLMs. Our code is available at https://github.com/StartluxLabs/Massive-Activations-HLA.

The Large Language Model from Power Law Decoder Representations (PLDR-LLM) and its attention, Power Law Graph Attention (PLGA), replace the fixed bilinear form of scaled dot-product attention (SDPA) with a learned, input-generated bilinear operator G_{LM}, built from a positive tensor A_{LM} by elementwise power laws. The architecture is fully specified, verified against pinned reference releases; claims are labeled theorem, conditional theorem, measurement, or conjecture. Unconditionally: PLGA contains SDPA exactly at G_{LM}=I; A_{LM} and A_P are strictly entrywise positive, with Perron-Frobenius structure on A_{LM}; the DAG regularizer has the NOTEARS walk-counting form and positivity obstructs exact acyclicity; and, under nonresonance (satisfied by standard rotary frequencies), a commutant criterion identifies which operators preserve relative-position dependence. An inference-collapse theorem: exact input invariance of deductive outputs collapses inference to generalized SDPA with a constant operator. Measured invariance: relative fluctuations of 10^{-6} and below; perturbation bounds quantify but do not certify cached inference; the assembled proxy misses the decoding margin. A conditional three-stage mechanism (rotary twirl, concentration, row-map contraction) is measured on a released checkpoint. Blockwise training and scoring under the global Gram are stated with explicit target exposure; on tested samples, block and sequential scoring select identical answers and agree on the published TruthfulQA probability-mass metric within 5times 10^{-5} per item. Self-organized criticality enters as a phenomenological framework with an intrinsic order parameter; open claims become falsifiable conjectures. Selected proof cores are machine-checked in Lean 4.

Test-time compute scaling is a primary driver of performance in large reasoning models (LRMs), but extreme inefficiency bounds current approaches, shifting the critical question from how much compute to spend, to where to allocate it. We formalize test-time reasoning as a constrained compute allocation problem over partial trajectories. Under a fixed hardware budget, existing paradigms fail to actively allocate the compute to the most promising partial progress: traditional parallel sampling treats traces independently and induces severe memory bottlenecks, while subtractive pruning starves hardware and fails to actively and sufficiently shift the output distribution. To overcome this dichotomy, we introduce Gambit, an inference algorithm that executes thought-level beam search. By periodically pruning unpromising trajectories and immediately branching from high-quality prefixes, Gambit dynamically concentrates compute onto the most promising reasoning traces via a light-weight scorer probing hidden states while maintaining continuous high hardware utilization. Extensive evaluations across multiple models and benchmarks demonstrate that Gambit strictly dominates existing baselines. Under identical hardware constraints, our method yields up to a +6.7\% absolute accuracy gain on HMMT-24 and +3.3\% on AIME-25 over pruning baselines, delivers >2times higher throughput on trace completion, and reduces total token consumption by up to 68.5\% relative to standard parallel sampling.

Spatial intelligence is becoming a foundation for embodied agents, robotic planning, and multimodal assistants. To improve the spatial reasoning ability of VLM agents, existing work has mainly followed two lines. One line uses post-training methods, such as supervised fine-tuning and reinforcement learning. Another line adopts an agentic paradigm in which the model calls external spatial tools, such as depth estimation and 3D reconstruction tools, to gather intermediate spatial evidence. We study a complementary and underexplored route: Can a frozen VLM agent improve its spatial reasoning through parameter-update-free self-evolution, without depending on external expert spatial tools at inference time? We present Spatial Memory Agent (SMA), an experience-grounded runtime framework that converts verified spatial experience into reusable transferable lessons. In a verifiable spatial environment, SMA queries the frozen VLM, obtains a predicted answer and reward, and uses verifier-guided reflection to distill compact transferable lessons from spatial experience. SMA further assigns each lesson a Transfer Reliability Score (TRS), which is initialized uniformly and calibrated from later retrieval outcomes as visit evidence of future transfer reliability. During read-only deployment, SMA retrieves lessons by semantic filter and similarity-TRS combined ranking, allowing the retrieved memory to guide frozen model inference. Across five representative spatial benchmarks and four base VLMs, SMA achieves the highest macro average in every base-model block and the best accuracy among the evaluated methods in most of the 20 evaluations, establishing a practical parameter-update-free path for spatial self-evolution across the evaluated frozen model scales and environments.

Recent advances in persistent personal-agent frameworks are making human-centered agent networks realistic deployment targets: each user can be served by an AI agent that acts on the user's behalf, maintains state, and communicates with other agents through social and task relations. In these networks, everyday tool use becomes multi-party owned-agent collaboration over personal workspaces, where files, records, tools, and policies are not directly visible across owners. Existing agent benchmarks study tool use and collaboration, but they do not provide an end-to-end sandbox for verifiable cross-user agent collaboration with realistic user digital workspaces or test how harmful actions can travel through the human-centered agent network. We introduce WeClawArena, an auditable benchmark and runtime sandbox for multi-party owned-agent collaboration over personal workspaces. WeClawArena targets collaborative tool-use tasks in which personal workspaces serve as both operational tools and personal constraints. The benchmark contains 124 base tasks across six cross-user task domains and expands them into 620 scenario variants, with one benign control and four attack-vector variants per base task. The sandbox records peer messages, tool calls, resource operations, governed decisions, and final workspace states. WeClawArena reports utility and attack success rate separately and audits attack success from bounded runtime evidence, supporting diagnosis of task breakdown, privacy leakage, poisoned evidence, and invalid authority paths.

Large-scale manipulation data is essential for robot learning, yet collecting robot demonstrations remains expensive and difficult to scale. Meanwhile, abundant egocentric human manipulation videos provide rich behavioral experiences, but transferring them across embodiments remains challenging due to differences between human hands and robotic end-effectors. Recent advances in video world models offer a promising pathway to synthesize robot-centric manipulation videos from human observations, while their cross-embodiment transfer capability remains largely unexplored. Therefore, we introduce H2R-Bench, a benchmark for evaluating cross-embodiment human-to-robot manipulation video generation, where models transform egocentric human demonstrations into robot manipulation videos under specified embodiments. Each benchmark instance contains a human demonstration video, target embodiment constraints, and source-grounded annotations covering task goals, action events, functional contacts, and object responses. H2R-Bench evaluates generated videos through five dimensions, including goal-state completion, action-event completion, functional contact transfer, embodiment correctness, and general video quality. We benchmark eleven state-of-the-art video generation models across six manipulation families and two robot embodiments. Our evaluation reveals that current video world models remain limited in human-to-robot manipulation transfer: even leading models often fail in embodiment consistency, functional interaction, and task execution. H2R-Bench provides a systematic diagnostic framework for evaluating whether video world models can bridge the human-to-robot embodiment gap and convert human manipulation observations into robot-centric training resources.

Large language model (LLM) agents integrated with external tools are vulnerable to indirect prompt injections embedded in environmental states. However, existing studies largely rely on manually implemented or reused environments, stochastic LLM-based tool simulation, and predefined injection locations, limiting scalable security research across broader domains. To bridge this gap, we propose **ToolHazard**, a scalable adversarial environment synthesis framework that reduces human engineering and supports expansion with additional seed domains and compute. Through an Environment Simulator, an Attacker Agent, and a User Simulator, ToolHazard synthesizes executable stateful environments, discovers viable injection points and generates environment-specific payloads, and constructs state-grounded long-horizon tasks. Based on ToolHazard, we build **ToolHazard-Bench** for stress-testing agents under complex workflows and diverse environmental attacks. Experiments reveal substantial agent vulnerabilities and show that injection timing and placement affect attack effectiveness. Moreover, ToolHazard-generated alignment data improves security on both ToolHazard-Bench and AgentDojo while preserving benign task utility.

Agent skills represent a standardized format for packaging procedural knowledge and domain expertise, serving within agent harness systems as an essential mechanism to continually constrain a language model's behavior space for repeatable, high-quality task execution. However, because strong closed-source models entail high inference costs, current popular agent harnesses, such as Codex and OpenClaw, remain prohibitively expensive when deploying these skills to accomplish real-world tasks. The rapid capability enhancement of open-source models deployable on consumer-grade GPUs presents a compelling opportunity to drastically reduce these costs by leveraging skill-based behavioral constraints. Nevertheless, automatically generating effective skills tailored specifically for such compact models remains a significant practical challenge. To address this, we propose SKILLER, a natural-language-driven reinforcement learning framework designed to automatically generate executor-specific skills for small models, which employs a strong model as the actor and critic, treats the small-model agent system as the environment, and propagates all reinforcement learning signals entirely via natural language. Extensive experimental evaluations across five relevant benchmarks using Qwen3.5-9B and Qwen3.5-4B demonstrate that SKILLER outperforms three open-source and one closed-source skill generation or evolution methods, achieving absolute gains ranging from 4.3 to 20.4 percentage points for the 9B model and 1.8 to 13.3 points for the 4B model, while remarkably matching the performance of strong closed-source models on single-skill tasks in SkillsBench. The project is available at https://github.com/DANG-ai/SKILLER.

Estimating human gaze targets from images in-the-wild is an important and formidable task. Existing approaches primarily employ brittle, multi-stage pipelines that require explicit inputs, like head bounding boxes and human pose, in order to identify the subject of gaze analysis. As a result, detection errors can cascade and lead to failure. Moreover, these prior works lack the flexibility of specifying the gaze analysis task via natural language prompting, an approach which has been shown to have significant benefits in convenience and scalability for other image analysis tasks. To overcome these limitations, we introduce the Promptable Gaze Target Estimation (PGE) task, a new end-to-end, concept-driven paradigm for gaze analysis. PGE conditions gaze prediction on flexible user text or visual prompts (e.g., "the boy in the red shirt" or "person in point [0.52, 0.48]") to identify a specific subject for gaze analysis. This approach integrates subject localization with gaze estimation, and eliminates the rigid dependency on intermediate analysis stages. We develop a scalable data engine to generate Gaze-Co (Gaze Estimation with Concepts), a dataset and benchmark of 120K high-quality, prompt-annotated image pairs. We also propose GazeAnywhere, the first model designed for PGE. GazeAnywhere uses a transformer-based detector to fuse features from frozen encoders and simultaneously solves subject localization, in/out-of-frame presence, and gaze target heatmap estimation. GazeAnywhere achieves state-of-the-art performance on multiple PGE benchmarks, setting a strong baseline for this new problem even on a difficult out-of-domain, real-world clinical dataset. GazeAnywhere is open-sourced in github.com/IrohXu/GazeAnywhere.

In evolutionary algorithms powered by language models, the LLM acts as a single operator that simultaneously updates structural components (like control flow) and continuous parameters. While LLMs can be good at the first, they are not efficient at the second, wasting tokens taking discrete jumps inside a trial and error loop. We resolve this by formalizing a hybrid nested search, in which an outer loop has the LLM propose a structural sketch, with numeric gaps, and an inner numerical optimizer tunes the sketch. Both the outer and inner solvers are pluggable: any text-based optimizer can be combined with a zero-order optimizer (CMA-ES), gradient-based routines, or MCMC samplers. We validate our framework across three scientific domains: (i) meta-optimizers on closed-form test functions, (ii) code-based policies for systems research and social dilemmas; and (iii) approximate Bayesian inference tasks. Across all three, the hybrid optimizer is superior to both vanilla LLM-driven search and pure numerical optimization baselines. Code at: https://github.com/vicgalle/hybrid-nested-search

论文研究有限练习预算下顺序任务中的机器人技能自主学习，并提出审慎练习算法，为能在预算内学会且可最大化预期累计回报的技能分配练习资源。该方法估计技能掌握时间和其解锁任务计划的累计回报，通过双线性规划精确计算预算最优分配；模拟与真实机器人实验显示，它能改进有限练习时间的使用和长时程规划。

艺术家条件图像生成不能只是在提示中添加艺术家姓名，因为模型可能依赖反复出现的母题、通用配色或过度代表的时代特征，而忽略用户想要的场景。本文提出 Atelier，通过显式控制状态、艺术家级知识、局部参考和全局及局部反馈来规划生成，并构建覆盖梵高与齐白石多种任务的 ArtIntentBench；作者称该方法提高了风格与结构保真度，同时减少了风格捷径替代。

灵巧操作需要操作者在整个任务过程中协调机械臂运动与精确、接触丰富的手指行为，因此完整一致的示范数据很难获得。嵌套策略学习系统让操作者控制机械臂，并通过单自由度离合器调节由视觉语言选择器激活的手部技能；这些示范随后训练一个部署时不依赖内层策略、同时控制机械臂和手部的外层视觉运动策略。真实世界实验表明，该方法提高了示范采集的可靠性与效率，并能支持自主策略学习。

论文介绍了梦想平台一代，这是一种面向机器人操作的动作条件视频世界模型，可根据观测帧、语言指令以及包含末端执行器位姿和夹爪状态的动作序列预测未来观测。该方法通过按机械臂注入三维刚体变换、加入深度分支和物体掩码约束几何与物体一致性，并将多步生成器蒸馏为少步学生模型以便部署；作者报告其在世界竞技场第二版挑战赛的两个赛道分别获得第一名和第二名。

该报告介绍了交互式长时域世界建模系统阿拉亚世界的改进版本，保持骨干网络、分块自回归生成方式和训练数据不变，重点重构条件信号的表示与融合。新版本以条件信号在潜表示和时间结构上贴近生成内容为原则，引入流式三维点缓存渲染、同一因果变分自编码器潜空间中的视觉条件，以及六项记忆和训练推理协议调整。

许多临床预测模型把术后结局视为从基线测量到未来终点的一步映射，但真实恢复过程包含异步记录的临床观测、用药变化、重复干预和生理测量，这些事件会持续改变风险评估。本文构建了一个干预感知的临床世界模型，将患者表示为结构化潜在状态，并按时间顺序更新术后事件；模型应用于房颤消融术后复发预测，在九十天恢复期内利用不规则记录预测长期复发风险，同时无需在推理时使用随访磁共振强度。

论文从因果视角研究世界模型，覆盖从感知观测、概念表示到描述环境动态结构的多个抽象层次。作者认为有用的世界模型不能只具备生成能力，还应捕捉并解释实体属性、实体间交互以及实体与环境的交互，并据此给出因果世界模型的形式定义，联系因果表示学习、物体中心学习、因果发现、结构因果模型和基于模型的决策，同时用可辨识性理论说明这些模型成分何时以及在多大等价范围内能够从数据恢复。

论文提出一个无需安装、基于浏览器的数字测试场，将选定海域转换为可复现的四维实验，覆盖任务配置、路线规划、洋流推进执行、观测生成和评分，并通过统一的计划到观测契约兼容手工路线、内置算法及外部经典或学习型规划器。受控观测系统模拟实验显示，五十四次任务均完成回收且没有硬约束违规，同时揭示了规划器排序和下潜策略中的运行与科学目标权衡；真实部署审计限定了当前运动学边界，源码锁定的滑翔机飞行模拟器还通过浏览器内运行实现了与原生版本的一致性。

临床智能需要从不完整观测中估计患者的潜在状况，因此论文将计算机断层扫描影像和临床语言视为共享患者状态的互补观测，并据此统一读出、重建与模拟任务。论文提出覆盖三类任务的计算机断层扫描基准，以及一个三十亿参数的多模态世界模型；该模型以共享变换器估计隐状态，生成状态查询答案、语言报告和条件性影像，并声称在三类任务上取得良好表现，同时通过临床结构化补全改善影像理解。

论文研究有限练习预算下序列任务的自主技能学习，提出一种主动技能学习算法，根据技能的掌握时间和可解锁任务计划的累积回报分配练习资源。该算法将原本涉及大量技能计划和长预算的组合优化转化为可由现成求解器精确计算的双线性规划，并在模拟和真实长时程操作任务中验证了其效果。

长时域手术机器人操作面临稀疏任务奖励和不规则交互变化，现有世界模型通常按原子步长进行想象，因而把可变时长的任务进展隐含在模型中。该方法从原子步长潜在轨迹学习稀疏事件证据，以协调事件级管理器和原子步长执行器，并通过事件转移模型预测下一事件边界的随机状态、片段持续时间和累计奖励，从而支持超出原子想象范围的可变时长规划；在基于模拟手术机器人环境的佩转任务上取得百分之九十八点七成功率。

接触丰富的操作失败通常要到机器人已经承诺接触后才被发现，腕部相机尤其容易因接近路径不佳而在检测器反应前推动、错过、滑动或扰动物体。该方法针对分块视觉运动策略建立接触前执行监控器：利用无标签机器人轨迹训练动作条件潜在世界模型预测多视角视觉嵌入，再用少量带标签的接触前片段训练失败探针，在即将接触时沿计划动作预测未来潜在状态，并在预示失败时中止；真实任务结果显示其预测优于直接预测和动作受扰动的消融方法，且无需修改底层策略即可用于在线机器人控制。

自动驾驶视觉语言动作模型能够进行端到端推理，但其计算成本过高，难以满足实时控制；论文将瓶颈归纳为视频帧视觉编码中的重复计算、语言模型预填充中的上下文重算、低熵推理令牌的串行生成，以及流匹配去噪中的均匀计算分配。该方法通过跨帧缓存、非自回归扩散草稿推测解码、自适应步缓存、图编译和内核融合协同降低开销，在十亿级参数驾驶模型上将延迟从七百一十七毫秒降至一百五十一毫秒，同时保持近似不变的精度并改善仿真闭环安全指标。

该方法针对组内相对策略优化在所有候选轨迹都失败时学习信号不足的问题，把低回报、低多样性的失败组转化为下一轮策略的特权监督：冻结轮次开始时的同规模模型作为教师，观察隐藏的未来轨迹，并沿学生生成前缀对答案词元进行自蒸馏，同时对所有轨迹组继续进行组相对策略优化。基于相同的视觉语言模型微调检查点，在六千零一十九个样本和一百五十个保留场景上，该方法将四条候选轨迹的平均位移误差从1.848米降至1.500米，并将持续失败比例从百分之十三点零三降至百分之十一点二零，改进主要来自减少少数严重失败轨迹。

自动驾驶规划需要同时利用语义约束和未来动力学预测，但视觉语言动作模型与世界动作模型通常分别侧重这两方面，直接通过词元级联合注意力融合会因语义捷径占据注意力而压制动力学预测。BrainWAM将二者转化为两条专门的动作通路，在紧凑动作表示层进行协调，并采用视频与动作解耦的异步整流流推理以缩短延迟；该方法在两个导航基准上分别取得八十九点五和八十九点六的分数，超过单独使用任一类模型的方法。

本文研究一种无需更新参数、也不依赖部署时外部空间专家工具的冻结视觉语言模型自我演化路径，并提出空间记忆智能体。该方法把经过验证的空间经验提炼为带迁移可靠性分数的可复用经验，在部署时通过联合检索记忆指导推理；在五个空间基准和四个基础模型上，它在各基础模型组中取得最高宏平均，并在二十次评测的大多数任务中达到最佳准确率。

视觉瓶颈通过把策略输入集中到相关区域，将“看哪里”和“如何行动”分开，有助于提高视觉运动学习的数据效率，但现有方法通常依赖空间标注或固定的动作派生裁剪。本文提出“搜寻器”，从冻结的视觉特征中迭代整合视觉证据，仅凭动作监督生成随任务和状态变化的区域，并将其用于图像裁剪、背景增强和点云过滤；仿真及真实机器人实验表明，该方法在域内以及光照和背景变化下都优于多种基线。

人形机器人动作跟踪是远程操控和全身模仿的基础，但现有逐帧姿态误差与人们对视频的感知常不一致，尤其会漏掉支撑不稳、足底滑动和落脚时机错误。该工作发布包含约一百五十三小时多名专业表演者光学动作轨迹的基准，覆盖四类动作并提供文本标签；同时利用包含两万四千段动作的一万二千对动作训练偏好对齐指标。实验表明，该指标比运动学指标更能预测人类偏好，并能发现接触和稳定性失败。