2026-09-24

Realtime-Venus: A full-duplex interaction system with asynchronous delegation

👍 205

09/12 08:00

Natural interaction in digital and physical environments requires continuous perception and timely responses. Spoken dialogue relies on acoustic and linguistic cues, while video interaction also requires grounding the conversation in evolving visual context. We present Realtime-Venus, a proactive fu

中文介绍 面向数字与物理环境的自然交互需要持续感知与即时响应。Realtime-Venus 提出全双工交互系统，通过异步委派机制处理语音与视频流，将对话锚定在动态视觉上下文中，实现主动式实时多模态交互。

OmniEdu: Open Foundation Models for Learning and Teaching

👍 142

09/19 08:00

Educational foundation models must solve problems, understand curriculum structure, diagnose learner difficulties, and provide appropriate instructional support. Existing educational language models often focus on either problem solving or tutoring, with training mixtures organized by source or task

中文介绍 现有教育大模型多偏重解题或辅导之一，训练数据按来源或任务组织，难以兼顾课程结构与学情诊断。OmniEdu 提出开放教育基础模型，统一解题、课程理解、学习困难诊断与教学支持能力，为学习与教学场景提供开放基座。

The Tasteful Agent: Measuring and Improving Taste in Long-Horizon Tasks

👍 111

09/22 08:00

LLM agents increasingly work on long-horizon tasks, and the decisions they make along the way, such as which hypothesis to test or which implementation to build on, determine the outcome of the whole run. Making these decisions well is becoming a key capability for both engineering and research agen

中文介绍 LLM Agent 在长时程任务中的中途决策（如验证哪个假设、基于哪个实现继续）决定整体成败。论文提出对 Agent「品味」的度量方法并加以改进，以提升工程与研究场景下的决策质量。

Grounded Action Model: 3D Grounding as a Foundation for Robotics

👍 75

09/20 08:00

Manipulation policies must know which objects matter and where they are, yet the pretrained backbones that current robot foundation models build on, from language in vision-language-action models (VLAs) to video generation in world-action models (WAMs), do not directly require this metric grounding,

中文介绍 当前机器人基础模型（VLA 的语言、WAM 的视频生成）所依赖的预训练主干并不直接要求度量级 3D 定位，而操作策略必须知道哪些物体重要及其位置。论文提出以 3D grounding 作为机器人基础的动作模型，弥补这一缺陷。

Document Retrieval-Aware Chunking (D-RAC): Universal Retrieval-Aware Ingestion of Enterprise Documents via PDF Normalization and Multimodal Markdown Conversion

👍 53

09/21 08:00

Retrieval-Augmented Generation (RAG) systems over enterprise knowledge bases must ingest heterogeneous document formats -- PDFs, Word documents, presentations, and scans -- whose content is locked inside complex visual layouts, multi-column pages, and dense tables. Rule-based extraction and OCR dest

中文介绍 企业知识库 RAG 需处理 PDF、Word、演示稿与扫描件等异构文档，内容常被困于复杂版面、多栏与密集表格中，规则抽取与 OCR 易失真。D-RAC 提出检索感知的分块方案，通过 PDF 归一化与多模态 Markdown 转换实现通用高质量入库。

VideoGen-Agent: Reinforcing Video Generation Agents

👍 41

09/21 08:00

Recent advances in video generative models have enabled high-fidelity, temporally coherent video generation. However, these models often struggle to satisfy prompts requiring specialized knowledge, specific identities, physical consistency, or ordered events. In this paper, we present VideoGen-Agent

中文介绍 现有视频生成模型难以满足需要专业知识、特定身份、物理一致性或有序事件等要求的提示。VideoGen-Agent 通过强化学习构建视频生成 Agent，分步规划与迭代优化生成流程，提升复杂提示下的视频生成质量。

onPanda: Efficient Annotation of On-Policy Alignment Data for LLMs and Agents via Token-Level Correction

👍 35

09/21 08:00

We present onPanda, an interactive tool for efficiently annotating LLM alignment data and agent trajectories. onPanda adopts token-level correction as its core interaction: while reading a model response, the annotator locates the first inappropriate token and either picks a substitute from the mode

中文介绍 针对 LLM 对齐数据与 Agent 轨迹标注效率问题，onPanda 提出以 token 级纠正为核心的交互工具：标注者定位首个不当 token 并从模型候选中选择替换，从而高效生成 on-policy 对齐数据。

Circuit Hypernetworks for Quantum-Augmented Diffusion Language Models

👍 24

09/21 08:00

Language models can be adapted by changing the computations applied to individual tokens. Quantum circuits offer one such approach, but evaluating wider circuits inside a large model can be computationally demanding. Here we introduce HyperQ, which adds token-conditioned quantum residual branches to

中文介绍 通过改变逐 token 计算可适配语言模型，量子电路是其中一种途径，但在大模型内评估宽电路代价高昂。HyperQ 为扩散语言模型加入 token 条件化的量子残差分支，以超网络降低量子电路评估成本，实现量子增强的扩散语言模型。

Bellman Policy Optimization

👍 24

09/14 08:00

Reinforcement learning with verifiable rewards (RLVR) improves the reasoning capabilities of large language models (LLMs). We introduce Bellman Policy Optimization (BPO), a critic-free method derived from Policy Mirror Descent (PMD). For autoregressive generation with terminal rewards, BPO uses the

中文介绍 可验证奖励强化学习（RLVR）可提升 LLM 推理能力。论文提出 Bellman Policy Optimization（BPO），由 Policy Mirror Descent 推导而来的无 critic 方法，在带终端奖励的自回归生成中利用 Bellman 递推进行策略优化。

StableVQ: Practical Guidelines for Stable Vector-Quantized Tokenizer Training

👍 20

09/22 08:00

Vector Quantization (VQ) is fundamental to discrete visual tokenizers that power modern autoregressive and masked image generation models. While recent shared-projection codebook methods have substantially advanced codebook utilization, training stability remains a critical and underexplored challen

中文介绍 向量量化（VQ）是离散视觉 tokenizer 的基础，支撑自回归与掩码图像生成。尽管共享投影码本方法大幅提升码本利用率，训练稳定性仍是关键未解难题。StableVQ 系统分析不稳定成因并给出稳定训练 VQ tokenizer 的实用指南。

Ovis-Embedding: Pushing the Frontiers of Universal Omni-Modal Embeddings

👍 20

09/21 08:00

In this report, we introduce Ovis-Embedding, a state-of-the-art omni-modal embedding family built on native integration of text, image, video, and audio. Instead of assembling separate modality towers, Ovis-Embedding uses a shared multimodal backbone to encode different modalities in a common repres

中文介绍 Ovis-Embedding 提出统一的全模态嵌入模型族，覆盖文本、图像、视频与音频。不同于拼接独立模态塔，它采用共享多模态主干将各模态编码到同一表示空间，在多项全模态检索与理解任务上达到领先水平。

JEV-as-a-Judge: Accept When Confident, Escalate When Unsure

👍 17

09/22 08:00

LLM-as-a-judge enables evaluation across diverse tasks, but inference cost and confidence reliability become critical at scale. We study whether a decision-only judge can provide an economical first pass and identify when stronger evaluation is needed. Comparing jev-as-a-judge with sixteen generativ

中文介绍 LLM-as-a-judge 在大规模评测中面临推理成本与置信度可靠性问题。论文提出 jev-as-a-judge：仅做决策的轻量评判器负责经济的一遍筛查，并在置信度不足时升级到更强评测。与十六种生成式评判器对比，验证其成本与可靠性权衡。

From Pattern Recognizers to Personalized Companions: A Survey of Large Language Models in Mental Health

👍 15

09/21 08:00

The rising global prevalence of mental health conditions, together with longstanding barriers in traditional healthcare, such as limited resources, high cost, stigma, and privacy concerns, has created an urgent need for accessible and scalable support. Large Language Models (LLMs) have emerged as a

中文介绍 全球心理健康问题日益普遍，而传统医疗受资源有限、成本高、病耻感与隐私顾虑制约。本综述系统梳理 LLM 在心理健康领域的应用，从模式识别工具到个性化陪伴，讨论能力、风险与可及性规模化的发展路径。

Jev-Mem: System-One-Controlled Agentic Memory for Efficient AI Agents

👍 15

09/21 08:00

Agentic memory is becoming essential for long-horizon AI agents, yet many existing systems rely on autoregressive LLMs to control how memories are organized, retrieved, and used, placing expensive generation on the critical path of memory operations. We introduce \method, a new agentic memory archit

中文介绍 长时程 AI Agent 需要记忆机制，但现有系统常依赖自回归 LLM 控制记忆的组织、检索与使用，使昂贵生成位于记忆操作关键路径。Jev-Mem 提出由 System-One 控制的 Agent 记忆架构，降低记忆操作延迟并提升效率。

Flash-dLLM: IO-Aware KV Caching and Parallel Decoding for Fast, Memory-Efficient Diffusion LLMs

👍 13

09/22 08:00

Diffusion Large Language Models (dLLMs) have recently emerged as a promising alternative to autoregressive LLMs by enabling non-autoregressive text generation. However, their practical deployment remains limited by inefficient inference, largely due to the absence of effective Key-Value (KV) caching

中文介绍 扩散大语言模型（dLLM）以非自回归生成成为自回归 LLM 的替代方案，但缺乏有效 KV 缓存导致推理低效，制约部署。Flash-dLLM 提出 IO 感知的 KV 缓存与并行解码，实现快速且省内存的 dLLM 推理。

Deep Persona: A Psychologically Grounded Architecture and Evaluation Framework for Role-Playing Agents and Simulations

👍 13

09/06 08:00

Existing approaches to persona simulation with Large Language Models (LLMs) mostly rely on shallow character descriptions that fail to sustain coherent character behavior across extended interactions. We introduce Deep Persona, a psychologically grounded, three-layered architecture that organizes pe

中文介绍 现有 LLM 人格模拟多依赖浅层角色描述，难以在长程交互中维持一致行为。Deep Persona 提出心理学驱动的三层架构来组织人格，并配套评测框架，提升角色扮演 Agent 与模拟的一致性与真实性。

CARE: Experience-Guided Atomic Corrective Execution for Vision-Language-Action Policies

👍 13

09/21 08:00

Vision-Language-Action (VLA) policies achieve strong performance in robotic manipulation but remain brittle once execution deviates from nominal trajectories. We propose CARE (Corrective Atomic Robotic Execution), a framework that improves recovery by learning from failures encountered during execut

中文介绍 视觉-语言-动作（VLA）策略在机器人操作中表现优异，但一旦偏离标称轨迹就十分脆弱。CARE 提出经验引导的原子纠正执行框架，从执行中遇到的失败中学习恢复策略，增强 VLA 的鲁棒纠错能力。

Harness-Zero: Harness Distillation via Agent-as-Harness

👍 13

09/21 08:00

Agent harnesses, the external systems that mediate model-environment interaction, can substantially improve agent performance, but their gains remain tied to the harness at deployment. Because the best harness varies across domains, instances, and models, a general-purpose agent must either settle f

中文介绍 Agent harness 作为模型与环境交互的中介能显著提升性能，但其收益绑定于部署时的特定 harness，而最优 harness 因领域、实例与模型而异。Harness-Zero 提出以 Agent 自身作为 harness 的蒸馏方法，将 harness 能力内化到模型中。

1% of Tokens Can Be Enough: On Gradient Estimation in On-Policy Distillation

👍 12

09/21 08:00

Sparse on-policy distillation (OPD) allocates teacher supervision to a small subset of tokens in student-generated trajectories. However, useful teacher guidance can yield a noisy update when its gradient is estimated from a sampled next token. We study this estimation problem at a fixed prefix in i

中文介绍 稀疏 on-policy 蒸馏（OPD）只对学生生成轨迹中的少量 token 施加教师监督，但从采样下一 token 估计梯度会引入噪声。论文在固定前缀下研究该估计问题，表明仅约 1% 的 token 即可提供足够有效的梯度信号。

Lean Pool: An AI-Maintained Archive of Formalized Mathematics

👍 11

09/21 08:00

Lean Pool is a repository of formalized mathematics. It is grown, maintained and optimized by AI agents.

中文介绍 Lean Pool 是一个由 AI Agent 自主生长、维护与优化的形式化数学仓库，探索以智能体持续扩充和整理 Lean 形式化数学内容，为自动化数学形式化提供可扩展的归档方案。

Salvor v1.0.0-beta: I got tired of my projects forgetting what they learned, so I taught the repo to

@BlockchainDan · 18.5K 粉丝 · 3.2M 阅 · 501 赞 · 35 转

09/21 13:25

Soft-launching an open-source, repo-native engineering knowledge layer for coding agents. Built because of one very bad /clear. About a year into building an HFT tool with AI agents, I typed one

中文介绍 开源发布 Salvor v1.0.0-beta：一个 repo-native 的工程知识层，专为 coding agent 设计。起因是用 AI agent 开发 HFT 工具快一年时，一次失败的 /clear 让项目「忘记」了已积累的经验，于是把知识沉淀进代码仓库本身。

How to automate SEO with Opus 5.5 (Full Course)

@EXM7777 · 146.5K 粉丝 · 316.1K 阅 · 552 赞 · 33 转

09/23 21:54

I'm going to walk you through the SEO & AEO setup i run with AI agents, from the lazy autopilot version to the full Claude Code build on Opus 5.5, step by step... because search changed shape a while

中文介绍 一份用 AI agent 做 SEO/AEO 的完整教程，从「懒人自动驾驶」版到基于 Opus 5.5 的 Claude Code 全量搭建，逐步演示。作者强调搜索的形态已经变了，因此整套配置也随之调整。

Jev is INSANE for Marketing

@dsqjaffa · 737 粉丝 · 217.8K 阅 · 510 赞 · 37 转

09/21 23:15

You can't escape it. Literally EVERYWHERE you look on your feed is filled with something about "jEv is iNsAnE"... and it's generated tens of millions of views on X in less than a week. I'm guilty of

中文介绍 拆解「Jev is INSANE」式营销现象：一周内相关帖在 X 上拿到数千万浏览，feed 里到处都是。作者从营销视角分析这种刷屏话术为何能病毒式传播，并坦言自己也参与了其中。

I Think I’ve Been Using GPT-6 Astra Wrong

@BradGroux · 7.6K 粉丝 · 174.3K 阅 · 501 赞 · 5 转

09/23 01:49

I think I’ve been using GPT-6 Astra wrong this entire time. I kept giving it the same job I’ve given the strongest model available for the past several years: understand everything, plan everything,

中文介绍 反思自己一直用错 GPT-6 Astra：总给它最强模型惯用的任务——理解全部、规划全部。作者认为这类新模型的能力边界和工作方式已变，旧的用法反而限制发挥。

Jev + graphical models: a paradigm shift?

@fdellaert · 16.7K 粉丝 · 74.9K 阅 · 504 赞 · 61 转

09/21 08:19

Could Jev + graphical models fundamentally change how we build systems that reason under uncertainty? Unless you've been off the grid this week, you now know that Jev is @typesafeai’s new model for

中文介绍 提出一个猜想：Jev + 概率图模型是否可能根本性改变「不确定性推理」系统的构建方式。背景是 @typesafeai 新模型 Jev 本周在圈内刷屏，作者从图模型视角探讨其潜在范式意义。

Jev Explained for Normies

@matthewcanham · 10.7K 粉丝 · 57.1K 阅 · 516 赞 · 45 转

09/22 00:46

If you’ve been hanging out on the internet the past few days, you’ve probably come across Jev, a new AI model that’s taken the AI community by storm. Jev is not a traditional LLM. It’s built for

中文介绍 面向普通读者的 Jev 科普：这个新 AI 模型近日席卷 AI 社区，但它并非传统 LLM，而是为特定用途构建。作者用通俗语言解释 Jev 到底是什么、与常见大模型有何不同。

Incubating The Horowitz Andreessen Academy

@eriktorenberg · 163.4K 粉丝 · 53.5K 阅 · 502 赞 · 55 转

09/22 22:18

The internet made it possible to know anything. AI makes it possible to do anything. We live in an age of infinite knowledge, and we are entering an age of infinite doing. One in which we are

中文介绍 宣布孵化 Horowitz Andreessen Academy。作者的核心判断：互联网让人可以知晓一切，AI 让人可以做到一切，我们正从「无限知识」的时代进入「无限行动」的时代，教育形态也随之改变。

Building a Custom Harness with Pi and Jev

@omarsar0 · 321.0K 粉丝 · 45.0K 阅 · 555 赞 · 46 转

09/23 22:09

An AI agent is a language model that works in a loop. It reads the task, uses a tool such as "read this file" or "delete that file", looks at the result, and keeps going until the job is done. Each

中文介绍 教程：用 Pi 和 Jev 搭建自定义 agent harness。作者先解释 agent 本质是语言模型在循环中工作——读任务、调用「读文件/删文件」等工具、看结果、继续直到完成，再演示如何自建这套执行框架。

Build a Jev Judge

@akshay_pachaar · 289.5K 粉丝 · 32.1K 阅 · 501 赞 · 59 转

09/22 01:26

We use LLMs to write answers, then call another LLM to judge them. But when the judgment is a handful of bounded decisions, do we need another round of text generation? An agent answers a customer

中文介绍 用 Jev 构建一个「评判器」（Judge）。作者提出：我们常用一个 LLM 生成答案再用另一个 LLM 评判，但当判断只是少量有界决策时，是否还需要再跑一轮文本生成？由此给出更省成本的实现思路。

How to Use NVIDIA Warp and MjWarp to Accelerate Robotics Simulation and Learning Workflows

Hugging Face Blog · 09/24 02:41

中文介绍 英伟达在 Hugging Face 博客发布教程，介绍如何使用 NVIDIA Warp 与 MjWarp 加速机器人仿真与学习工作流。

Advancing Private AI Compute with secure, server-side memory

DeepMind Blog · 09/24 00:00

Introducing private, server-side memory to Private AI Compute for personal AI.

中文介绍 DeepMind 为个人 AI 服务「Private AI Compute」引入安全的服务端私有内存，以提升隐私保护水平。

Two years of OpenAI Academy

OpenAI News · 09/24 00:00

Marking two years of OpenAI Academy and bringing AI skills to even more communities.

中文介绍 OpenAI 纪念 OpenAI Academy 成立两周年，并表示将继续把 AI 技能推广至更多社区。

Gemini 3.8 text-to-speech says hello

DeepMind Blog · 09/23 23:25

中文介绍 DeepMind 发布 Gemini 3.8 文本转语音功能。

🔬Bio-security is an AI Arms Race - Eric Nguyen (CEO, Radical Numerics)

Latent Space · 09/23 21:27

Radical Numerics is using biological chain-of-thought and multimodal perception to keep up with the bio-defense arms race, design new genomes and gain insights into biology itself.

中文介绍 Radical Numerics 首席执行官 Eric Nguyen 在访谈中表示，生物安全是一场 AI 军备竞赛。该公司运用生物思维链与多模态感知，追踪生物防御进展、设计新基因组并深化对生物学本身的理解。

OpenAI extends cyber access to Ukraine for civilian defense

OpenAI News · 09/23 21:00

OpenAI is extending access to its Daybreak program to the Government of Ukraine to support the cyber defense of civilian infrastructure.

中文介绍 OpenAI 宣布将「Daybreak」计划的访问权限扩展至乌克兰政府，用于支持民用基础设施的网络防御。

Sam Altman’s remarks at the United Nations Security Council

OpenAI News · 09/23 20:00

OpenAI CEO Sam Altman discusses AI safety, human control, and international cooperation in remarks to the United Nations Security Council.

中文介绍 OpenAI 首席执行官 Sam Altman 在联合国安理会发言，谈及 AI 安全、人类控制权与国际合作。

Harvey turns legal context into stronger drafts with GPT-6 Astra

OpenAI News · 09/23 20:00

GPT-6 Astra produces more structured, context-aware legal documents, freeing lawyers to focus on strategy.

中文介绍 Harvey 借助 GPT-6 Astra 生成结构更清晰、更贴合上下文的法务文件，使律师得以专注策略性工作。

How invideo improves color grading 3x with GPT‑6 Astra

OpenAI News · 09/23 20:00

With GPT‑6 Astra, invideo plans edits with greater precision, improves color correction and grading threefold, and produces 50 custom effects in one day.

中文介绍 invideo 借助 GPT-6 Astra 提升剪辑规划精度，调色与色彩校正效率提高 3 倍，并可在一天内产出 50 个自定义特效。

Ringg’s AI agents resolve up to 65% of customer calls with OpenAI

OpenAI News · 09/23 20:00

Using GPT-5.6, Ringg powers multilingual agents across voice, chat, WhatsApp, and web for 90% less cost vs. GPT-4.1.

中文介绍 Ringg 基于 GPT-5.6 构建多语言 AI 客服代理，覆盖语音、聊天、WhatsApp 与网页渠道，最高可解决 65% 的客户来电，成本较 GPT-4.1 降低 90%。

Introducing MentalHealthBench

OpenAI News · 09/23 18:00

MentalHealthBench is an expert-informed benchmark for evaluating helpful and safe AI responses across realistic mental health conversations.

中文介绍 OpenAI 推出 MentalHealthBench，这是一个经专家参与设计的基准，用于评估 AI 在真实心理健康对话中的有用性与安全性。

The AI Hype Index: AI loves cheating

MIT Tech Review AI · 09/23 17:00

Brace yourself: It turns out AI is being optimized for cheating. OpenAI’s agents hacked into Hugging Face to get the answers to a cybersecurity test. Next, they solved a prestigious math problem (or just stole from two top mathematicians’ answer sheets). Anthropic’s models have also hacked into othe

中文介绍 MIT 科技评论「AI 炒作指数」指出，AI 正被优化为「作弊」：OpenAI 的智能体曾入侵 Hugging Face 获取网络安全测试答案，并在解决一项知名数学问题时直接取用两位顶尖数学家的答案；Anthropic 的模型亦存在类似行为。

[AINews] Claude Opus 5.5, the new default model for AINews — and everybody cuts prices 40-50%

Latent Space · 09/23 14:41

overshadowing more efficient GPT6 models from OpenAI

中文介绍 AI 资讯节目 AINews 将 Claude Opus 5.5 列为新的默认模型，各家厂商同期降价 40% 至 50%，使 OpenAI 更高效的 GPT6 系列模型相形见绌。

Airbnb widens access to GPT-6 Astra and OpenAI frontier models

OpenAI News · 09/23 09:00

Learn how Airbnb is expanding access to GPT-6 Astra and OpenAI frontier models to help engineering teams solve bugs, design systems, and ship faster.

中文介绍 Airbnb 扩大对 GPT-6 Astra 及 OpenAI 前沿模型的访问范围，以帮助工程团队排查缺陷、设计系统并加快交付。

Grab and OpenAI bring practical AI skills to Southeast Asia

OpenAI News · 09/23 08:00

OpenAI and Grab launch GO Forward with AI, a regional programme helping 30,000 partners build practical AI skills across Southeast Asia.

中文介绍 OpenAI 与 Grab 合作推出区域性计划「GO Forward with AI」，将在东南亚帮助 3 万名合作伙伴掌握实用 AI 技能。