2026-07-19

LongStraw: Long-Context RL Beyond 2M Tokens under a Fixed GPU Budget

👍 172

07/16 08:00

A growing gap separates inference context lengths from RL post-training: inference systems are approaching million-token contexts, while post-training workloads often remain at 256K tokens or below and rely on length generalization at deployment. The gap is especially important for AI agents, whose

Boogu-Image-0.1: Boosting Open-Source Unified Multimodal Understanding and Generation

👍 125

07/14 08:00

We introduce Boogu-Image-0.1, an open-source unified multimodal understanding and generation model family, comprising Base, Turbo, Edit, and Edit-Turbo variants. It delivers competitive performance in high-quality text-to-image generation, fast inference, instruction-based editing, and bilingual (Ch

中文介绍 开源统一多模态理解与生成模型家族 Boogu-Image-0.1，包含 Base、Turbo、Edit、Edit-Turbo 变体，在文本到图像生成、快速推理、基于指令的编辑及中英双语场景中取得竞争性能。

VideoChat3: Fully Open Video MLLM for Efficient and Generalist Video Understanding

👍 120

07/16 08:00

Recent advances in video understanding have spanned motion, long video, and streaming interaction, driving this field toward real-world applications. Despite this progress, current open-source models remain limited in several ways. They often struggle to generalize across diverse video types, making

中文介绍 提出全开源视频多模态大模型 VideoChat3，实现高效且通用的视频理解，覆盖运动、长视频和流式交互，克服现有模型在多种视频类型上泛化不足的局限。

Ring-Zero: Scaling Zero RL to a Trillion Parameters for Emergent Reasoning

👍 90

07/14 08:00

Reinforcement learning with verifiable rewards without human-annotated data, often referred to as zero RL, has emerged as a powerful paradigm for eliciting chain-of-thought reasoning. However, due to computational constraints, existing studies are largely restricted to small models, leaving the trai

中文介绍 将零 RL（基于可验证奖励的强化学习）扩展至万亿参数规模，用于涌现 CoT 推理，突破以往计算约束仅限小模型的局限，展示大规模零 RL 的可行性。

SEED: Self-Evolving On-Policy Distillation for Agentic Reinforcement Learning

👍 82

07/16 08:00

Large language models are increasingly trained as interactive agents for long-horizon tasks involving multi-turn interaction, tool use, and environment feedback. Outcome-based reinforcement learning (RL) provides a practical optimization paradigm, but its sparse trajectory-level rewards offer limite

中文介绍 针对长程任务智能体 RL 中稀疏轨迹奖励问题，提出自进化在线策略蒸馏方法 SEED，通过自改进机制提升多轮交互和工具使用场景下的学习效率。

SearchOS-V1: Towards Robust Open-Domain Information-Seeking Agent Collaboration

👍 58

07/16 08:00

Recent advances in Tool-Integrated Large Language Models have made web search a core capability of information-seeking agents. However, as interaction histories grow, agents increasingly struggle to track task progress. When search attempts fail to yield useful evidence, current single- and multi-ag

中文介绍 提出 SearchOS-V1，用于开放域信息搜索代理的鲁棒协作，解决交互历史增长导致的任务跟踪困难，通过单/多代理协作框架提升搜索证据收集效率。

BadWAM: When World-Action Models Dream Right but Act Wrong

👍 44

07/16 08:00

World-action models (WAMs) are emerging as a promising foundation for embodied control: rather than predicting actions alone, they learn representations that couple action generation with future world prediction. This coupling is often viewed as a source of robustness, interpretability, and safety,

中文介绍 揭示世界动作模型（WAM）中「预测正确但动作失败」的矛盾，指出未来世界预测与动作生成之间的耦合可能引入安全与鲁棒性问题，并分析其根源。

AgentCompass: A Unified Evaluation Infrastructure for Agent Capabilities

👍 38

07/15 08:00

As Large Language Models (LLMs) evolve into autonomous agents, the need for unified evaluation infrastructure becomes critical. However, current evaluation pipelines remain highly fragmented and tightly coupled, hindering reproducibility and causing redundant engineering. To address this, we introdu

中文介绍 提出统一智能体能力评估基础设施 AgentCompass，解决当前评估管线碎片化、耦合度高的问题，提升可复现性并减少重复工程。

KeyFrame-Compass: Towards Comprehensive Evaluation of Keyframe-Conditioned Video Generation

👍 35

07/15 08:00

Video generation increasingly relies on keyframe-based workflows, where creators specify a sequence of reference images to guide generation. Although recent models support multi-keyframe conditioning, it remains unclear whether they can faithfully reproduce the prescribed keyframes while maintaining

中文介绍 构建全面评估关键帧条件视频生成的基准 KeyFrame-Compass，衡量模型是否忠实再现参考帧同时保持时间连贯性，揭示现有模型在该任务上的不足。

PolicyShiftGuard: Benchmarking and Improving Policy-Adaptive Image Guardrails

👍 35

07/07 08:00

Image guardrails are typically trained and evaluated under a fixed safety policy, implicitly treating safety as an intrinsic property of an image. Real deployments are different: the same image may be allowed in one product, restricted in another, and newly disallowed when a policy boundary changes.

中文介绍 提出 PolicyShiftGuard 基准，评估和改进图像安全护栏在动态政策下的自适应能力，解决同一图像在不同产品中安全属性变化的问题。

MetaView: Monocular Novel View Synthesis with Scale-Aware Implicit Geometry Priors

👍 33

07/13 08:00

Current visual generation models are capable of producing high-quality content, yet they lack a coherent perception of the spatial structure. Existing generative novel view synthesis methods typically introduce explicit geometry priors, which enforce spatial consistency but inherently restrict gener

中文介绍 提出 MetaView，单目新视角合成方法，引入尺度感知的隐式几何先验，在维持空间一致性的同时避免限制生成多样性。

MultiRef-Compass: Towards Comprehensive Evaluation of Multi-Reference-to-Audio-Video Generation

👍 31

07/15 08:00

Multi-reference-to-audio-video (MR2AV) generation aims to generate coherent audio-video content conditioned on multiple references and textual instructions. Existing benchmarks mainly focus on text-driven generation, single-reference subject preservation, or isolated audio-video alignment, leaving t

中文介绍 构建多参考条件音频-视频生成评估基准 MultiRef-Compass，填补现有基准仅关注文本驱动或单参考的空白，推动多模态内容生成标准化评测。

UniVR: Thinking in Visual Space for Unified Visual Reasoning

👍 28

07/14 08:00

Learning broad world knowledge directly from raw visual data is a fundamental capability of intelligence. We introduce UniVR, the first investigation into simultaneously learning complex reasoning, fine-grained physical dynamics, and long-term planning from pure visual demonstrations. At its core, U

中文介绍 首次探索从纯视觉演示中同时学习复杂推理、物理动态和长期规划，提出 UniVR 框架，以视觉空间为统一表征进行端到端推理。

Concurrent Image Understanding and Generation: Self-Correcting Coupled Markov Jump Processes

👍 27

07/14 08:00

Human cognition does not separate understanding and generation. A teacher at a whiteboard speaks and draws together, each modality reshapes the other. In this paper, we bring this coupled loop to artificial systems. Masked Diffusion Models (MDMs) are ideally suited to this task, yet existing sampler

中文介绍 模拟人类认知中理解与生成相互塑造的过程，提出自校正耦合马尔可夫跳跃过程，使掩码扩散模型同时执行图像理解与生成。

GigaWorld-Policy-0.5: A Faster and Stronger WAM Empowered by AutoResearch

👍 26

07/15 08:00

World Action Models (WAMs) improve robot policy learning by jointly modeling actions and future visual observations, using future scene evolution as dense supervision for physically grounded action generation. However, a common design in existing WAMs is to explicitly generate future videos at infer

中文介绍 提出 GigaWorld-Policy-0.5，加速并增强世界动作模型，通过 AutoResearch 自动研究策略避免推理时显式生成未来视频，提升物理动作生成效率。

Self-Improvements in Modern Agentic Systems: A Survey

👍 23

07/14 08:00

Self-improving autonomous agents are moving from research prototypes to deployed systems. The primary goal is controllable evolution, or adaptation, from experience with minimal or even no human input. This survey frames modern self-improving agents as adaptive systems that convert experience into a

中文介绍 综述现代自主智能体系统中的自我改进方法，将智能体视为将经验转化为能力的自适应系统，系统整理可控演化路径与关键挑战。

Spectral Rewiring for Exploration, Purification, and Model Merging

👍 22

07/03 08:00

Reinforcement learning has become a standard post-training recipe for large language models, but dense full-parameter updates create two deployment-relevant bottlenecks: suppressed reasoning performance, often reflected by premature saturation of test-time scaling, and interference when consolidatin

中文介绍 提出谱域重连方法，解决 RL 后训练中全参数更新导致的推理性能饱和与模型合并干扰，实现探索、净化与模型合并的统一框架。

RxBrain: Embodied Cognition Foundation Model with Joint Language-Visual Reasoning and Imagination

👍 21

07/15 08:00

Embodied cognition requires agents to connect high-level task reasoning with the physical states to be achieved. We introduce Hy-Embodied-RxBrain, an embodied cognition foundation model with joint language-visual reasoning and imagination. Unlike vision-language models that emphasize scene understan

中文介绍 提出具身认知基础模型 RxBrain，联合语言-视觉推理与想象，连接高层任务推理与物理状态达成，超越仅强调场景理解的视觉-语言模型。

Hallo4D: Multi-Modal Hallucination Mitigation for Consistent Spatio-Temporal Generation

👍 18

07/15 08:00

While recent advances in 3D generation have enabled impressive visual synthesis, existing methods often rely on 2D diffusion supervision without explicit mechanisms for geometric consistency, leading to spatial hallucinations such as duplicated structures and misaligned geometry. These issues become

中文介绍 提出多模态幻觉缓解方法 Hallo4D，用于一致时空生成，解决 3D 生成中因缺乏几何约束导致的重复结构和几何错位问题。

Demystifying On-Policy Distillation: Roles, Pathologies, and Regulations

👍 17

07/15 08:00

On-policy distillation (OPD) has become a key paradigm in LLM post-training, yet its training dynamics remain poorly understood. We present a systematic study examining the role, pathologies, and regulations of OPD. We first clarify the role of OPD as an exploration catalyst: it steers the student t

中文介绍 系统研究在线策略蒸馏（OPD）在 LLM 后训练中的角色、病理与调控机制，阐明 OPD 作为探索催化剂的本质，并提出优化策略。

GLM 5.2 With Vision

@part_harry_ · 1.5K 粉丝 · 225.6K 阅 · 503 赞 · 46 转

07/16 12:24

GLM 5.2 is one of the best currently available open source language models. However, unlike other flagship models like Qwen, Kimi and Minimax, GLM 5.2 does not support image inputs. We took this as a

中文介绍 为 GLM 5.2 开源模型添加视觉输入支持的方法，弥补了其不支持图像输入、落后于 Qwen、Kimi 等模型的短板，展示了如何将纯语言模型扩展为多模态模型。

The Self-Driving Company

@amasad · 472.6K 粉丝 · 223.7K 阅 · 560 赞 · 65 转

07/17 01:07

We are beginning to see what happens when a company learns to operate itself. In the past six months, engineers at Replit have nearly tripled code output. Review times held steady. Reversions and

中文介绍 Replit 工程师借助 AI 在半年内将近代码产出提升三倍，同时代码审查时间保持稳定，展示了 AI 驱动公司「自我运营」的高效实践，为工程效率提升提供真实案例。

The Next Horizon in Agents

@btaylor · 158.6K 粉丝 · 144.3K 阅 · 511 赞 · 40 转

07/16 22:48

Introducing Horizon Today, Sierra is announcing Horizon, a platform that enables agents to pursue long-horizon goals, like originating a loan or getting prior authorization for a healthcare procedure.

中文介绍 Sierra 推出 Horizon 平台，使 AI 代理能够执行贷款审批、医疗预授权等长周期任务，拓展了代理的能力边界，从短时交互走向长期目标导向。

Gemini API Managed Agents: free tier, triggers & more

@GoogleAIStudio · 186.6K 粉丝 · 111.0K 阅 · 512 赞 · 51 转

07/17 01:05

Today we’re announcing new capabilities for Managed Agents in Gemini API, including free tier availability, budget control guardrails, and scheduled triggers. Building on our previous release of

中文介绍 Google AI Studio 宣布 Gemini API Managed Agents 新增免费层、预算控制护栏和定时触发器，降低开发门槛并提升运维灵活性，方便代理的批量部署与调度。

How we built our knowledge base

@cerebras · 60.4K 粉丝 · 70.7K 阅 · 509 赞 · 38 转

07/17 02:27

Authors: @hi_im_isaac_, @learnwdaniel, @gaozenghao note: the interactive version of full technical blog available: https://www.cerebras.ai/blog/how-we-built-our-knowledge-base Employees ask our internal knowledge base more than 15,000

中文介绍 Cerebras 分享其内部知识库的构建技术细节，该库每日被员工查询超过 15,000 次，涉及索引、检索等工程实践，为企业级知识库建设提供参考。

Kimi K3 on KernelBench

@elliotarledge · 40.7K 粉丝 · 69.3K 阅 · 506 赞 · 45 转

07/17 17:24

This is the Kimi K3 post you guys have been waiting for. I got some early access to this model and have been testing it on kernels, and even before seeing the benchmark scores I was impressed with its

中文介绍 早期评测 Kimi K3 模型在 KernelBench 上的表现，其在 GPU kernel 编写与优化方面表现出色，显示出该模型在底层计算任务上的潜力，值得关注。

From Loop Engineering to Graph Engineering?

@IntuitMachine · 62.3K 粉丝 · 50.7K 阅 · 552 赞 · 47 转

07/18 18:00

What the shift in AI agent architecture is really about Peter Steinberger just posted nine words that gathered thousands of likes: https://x.com/steipete/status/2078277297791189132 "Are we still talking loops or did we

中文介绍 探讨 AI 代理架构从「循环工程」向「图工程」的转变，借 Peter Steinberger 的观点指出代理设计正从线性调用走向图结构编排，引领下一代架构讨论。

[AINews] not much happened today

Latent Space · 07/18 12:30

a quiet day

中文介绍 当日AI领域消息平淡，暂无重大事件发生。

Fine-tune video and image models at scale with NVIDIA NeMo Automodel and 🤗 Diffusers

Hugging Face Blog · 07/17 23:57

中文介绍 NVIDIA与Hugging Face合作，利用NeMo Automodel和Diffusers实现大规模微调视频和图像模型。

A scorecard for the AI age

OpenAI News · 07/17 18:00

Sarah Friar, CFO of OpenAI, introduces a practical AI scorecard to measure ROI through useful work, cost per successful task, dependability, and return on compute.

中文介绍 OpenAI首席财务官Sarah Friar推出AI评分卡，通过有用工作、单次任务成本、可靠性和计算回报衡量AI投资回报率。

The risk of weather data sabotage is rising

MIT Tech Review AI · 07/17 16:57

Every morning, airline dispatchers, grid operators, and farmers around the world make decisions based on the same thing: a weather forecast. While these forecasts are something that most people glance at for two seconds, weather predictions influence major strategic decisions in many industries, wit

中文介绍 天气数据破坏风险日益增加，天气预报影响航空调度、电网运营和农业决策，安全威胁值得关注。

not much happened today

Smol AI News · 07/17 13:44

**Moonshot's Kimi K3 release** has sparked a reassessment of **Chinese open-weight models**' proximity to the frontier, with strong performance in coding, agentic tasks, and long-horizon knowledge work. The strategic focus has shifted from a "compute moat" to an "efficiency stack" involving **MoE ro

中文介绍 Moonshot发布Kimi K3开源模型，性能接近前沿，在编程、代理任务和长周期知识工作中表现优异，业界焦点从算力转向效率堆栈。

[AINews] Kimi K3 2.8T-A50B: the largest open model ever released; Opus 4.8-class at Sonnet 5 pricing

Latent Space · 07/17 09:46

a great week for open models continues.

中文介绍 Kimi K3模型参数达2.8T，采用A50B架构，是史上最大开源模型，性能媲美Opus 4.8，定价与Sonnet 5相近。

Kimi K3 🌕, Gemini 3.5 delayed ⏳, crushing ARC-AGI 3 🤖

TLDR AI · 07/17 08:00

中文介绍 当日AI新闻：Kimi K3发布，Gemini 3.5推迟，ARC-AGI 3被攻克。

Why teens deserve access to safe AI

OpenAI News · 07/17 00:00

Learn how OpenAI is making ChatGPT safer for teens with age-appropriate protections, learning tools, parental controls, and expert partnerships.

中文介绍 OpenAI致力于让ChatGPT对青少年更安全，提供适龄保护、学习工具、家长控制并与专家合作。

🔬 The Lab of the Future Should Feel Like a Data Center — Andy Beam & Rafa Gómez-Bombarelli, Lila Sciences

Latent Space · 07/16 21:30

Lila is betting that science, not the internet, is the last untapped source of training data. We went to find out what that actually looks like in a room full of robots.

中文介绍 Lila Sciences认为科学是最后未被充分利用的训练数据源，实验室应像数据中心，研究团队在充满机器人的环境中探索。

Newer Models, Same Advantage

Hugging Face Blog · 07/16 19:49

中文介绍 本文探讨新模型在保持原有优势方面的表现，强调模型迭代中的持续优势。

Our approach to bioresilience

DeepMind Blog · 07/16 17:30

Google DeepMind and Isomorphic Labs are sharing our joint approach to bioresilience and AI models.

中文介绍 Google DeepMind与Isomorphic Labs公布其生物弹性联合方法，涉及AI模型在生物安全领域的应用。

[AINews] Thinky's Inkling: 975B-A41B multimodal, new best American Apache 2.0 open model (with Inkling-Small, 276B-A12B)

Latent Space · 07/16 14:18

Thinky's first full LLM release is a banger and bonus: it's open weights!

中文介绍 Thinky发布首个完整大语言模型Inkling，975B参数A41B多模态，提供开源权重，另有小模型Inkling-Small（276B-A12B），成为美国最佳Apache 2.0开源模型。

not much happened today

Smol AI News · 07/16 13:44

**Moonshot AI** launched **Kimi K3**, a frontier-class open-weights model with **2.8T parameters**, **1M-token context window**, and **native multimodal input**. It features novel **Kimi Delta Attention (KDA)** enabling up to **6.3x faster decoding** and **Attention Residuals** for **~25% higher tra

中文介绍 Moonshot AI发布Kimi K3开源模型，拥有2.8T参数、1M token上下文窗口、原生多模态输入，采用Kimi Delta Attention实现最高6.3倍解码加速。

How Cars24 scales conversations and builds faster with OpenAI

OpenAI News · 07/16 08:00

Cars24 uses OpenAI-powered voice and chat agents to handle 1M+ monthly conversation minutes, recover 12% of lost leads, and bring agentic workflows to teams across the company.

中文介绍 汽车交易平台Cars24利用OpenAI驱动的语音和聊天代理，每月处理超100万分钟对话，挽回12%流失线索，并在全公司推广代理工作流。

Security incident disclosure — July 2026

Hugging Face Blog · 07/16 08:00

中文介绍 Hugging Face披露2026年7月安全事件，并公布应对措施，提醒用户关注。