2026-10-23

FrontierCS 2: A Benchmark for Computer Science R&D

👍 2

10/21 08:00

Can LLM agents make contributions to frontier computer science R&D? We introduce FrontierCS 2, a benchmark of 48 open-ended tasks. We take a paper and the code its authors released and delete the paper's contribution. The deleted code becomes the human reference that sets the bar. Earlier open-ended

Looped Uniform Diffusion Language Models

👍 4

10/21 08:00

Uniform diffusion language models (UDLMs) generate text by iteratively denoising sequences of ordinary vocabulary tokens, any of which may be corrupted and subsequently revised. During training, UDLMs must denoise from a context that may itself be unreliable, whereas masked diffusion language models

Quant-Arena: Benchmarking LLM Agents for Quantitative Trading Research

👍 21

10/21 08:00

Quantitative research is an iterative loop: a researcher proposes a market hypothesis, implements it as an executable signal, tests it on historical data, and decides what to keep. Existing benchmarks score formulas, code, or individual trading decisions, leaving open whether large language models (

ToM2Vec: Continuous Theory-of-Mind Tokens for Multi-Agent Coordination

👍 7

10/21 08:00

Forming an accurate theory-of-mind (ToM) is key to successful multi-agent collaboration and requires reasoning under uncertainty to capture multiple possible partner states. This is hard to do in the text representation space native to large language model (LLM) agents, which more easily describe a

Retrieval Augmented Generation Compositionality

👍 26

10/21 08:00

Retrieval Augmented Generation (RAG) grounds large language models in external knowledge. Modern pipelines combine embedding, chunking, retrieval, reranking, query augmentation, context compression, and agentic verification, yet these techniques are typically evaluated one at a time. It remains uncl

From VLM Planners to Robotic Agents: Self-Evolving Harnesses for History-Dependent Manipulation

👍 11

10/21 08:00

Vision-language-action (VLA) models achieve strong performance across embodied manipulation tasks by unifying visual understanding, language conditioning, and action generation. However, long-horizon tasks can involve non-Markovian dependencies: successful decisions may require retaining task progre

PACE: Periodic and Channel-Aligned Encoding of Four-Bit Recurrent States

👍 19

10/21 08:00

Linear-attention language models keep their context in a recurrent state of fixed size, where an inference system stores one copy of this state for every sequence it decodes at the same time. Storing the state at lower precision therefore lets more sequences share an accelerator. Existing state quan

Think Harder Where It Matters: Item-Level Adaptive Recursion for Sequential Recommendation

👍 11

10/21 08:00

Existing sequential recommenders usually allocate a fixed computational budget to every item in a user's interaction history, regardless of the item's contribution to the final prediction. We challenge this de facto design with two diagnostic findings. (1) Non-uniform computational need: items withi

Multimodal Embodied Agents: A Unified Capability-centric Perspective from Computer-Use to Robot-Use

👍 18

10/21 08:00

Multimodal agents (MMAs) sustain interaction through reasoning, memory, tools, and feedback, most visibly as computer-use agents, while robotic systems (RSs) couple sensing and actuation under physical dynamics. We define multimodal embodied agents (MMEAs) as goal-directed systems that couple multim

SpatialBrain: Spatial Intelligence Growing from Experience

👍 12

10/21 08:00

Spatial reasoning requires more than access to geometric tools: a reasoner must decide when direct observation is sufficient and when to reconstruct, measure, or change reference frames. We present SpatialBrain, a framework that retains experience about these decisions while keeping its vision-langu

TraceForge: Distilling with Your Own Thinking

👍 19

10/21 08:00

Using an LLM to generate training data for another is among the most general ways to train one, and it remains common practice. Some frontier teams have closed that route for their own models: what comes back is an answer, or a short summary, without the detailed reasoning trace behind it. That is a

Evolving with Open Worlds: Towards Recursive Self-Improvement for Future Forecasting

👍 11

10/21 08:00

Most contemporary artificial intelligence (AI) systems are released with frozen parameters and improve through external updates, whereas an ideal system would keep improving from its own experience, a capacity we call recursive self-improvement (RSI). Yet studying RSI requires an environment that is

Agentic World Action Models

👍 11

10/21 08:00

Recent world action models provide visual foresight and future prediction for robot manipulation, but lack explicit mechanisms for compositional skill reuse. Agentic robotics combines code-based planning and reasoning with reusable skills, but often lacks predictive grounding in physical dynamics. W

Hidden Messages Between Diffusion Language Model Agents: Causal Evidence for Latent Communication

👍 7

10/20 08:00

Diffusion language models form answers through a sequence of revisable continuous states, but multi-agent systems usually discard those states and make agents communicate only after decoding text. We ask whether one independent diffusion trajectory can pass a compact hidden message into another and,

DiffResearch: Accelerating Deep Research with Diffusion Language Models

👍 6

10/20 08:00

Deep research systems iteratively plan, gather evidence, synthesize, and revise, making end-to-end latency a key bottleneck for interactive use. Almost all production systems are built on autoregressive (AR) language models and inherit left-to-right decoding in planning and long-form writing. We pre

Automating ML for Science: Can Frontier Agents Climb Scientific Hills in the Wild?

👍 13

10/20 08:00

Frontier coding agents are increasingly framed as end-to-end automated researchers, and ML for science offers a demanding test of that promise: improving a scientific model often requires domain insight beyond routine ML engineering. We introduce Agent4Sci, a long-horizon testbed for measuring how f

Scaling in Games: Continued Pre-Training For Embodied Agents In Diverse Virtual Worlds

👍 46

10/20 08:00

Scaling data, diversity and model capacity has improved embodied learning in physical environments, where diverse experiences remain grounded in shared physical laws. Whether these gains extend to virtual worlds with different visual styles, dynamics, controls, and objectives remains unclear. We inv

Building AI Research Teams for Autonomous Discovery of Linear Attention Architectures

👍 16

10/20 08:00

Autonomous agents can increasingly propose ideas, modify code, and run experiments, creating an opportunity to accelerate scientific discovery using teams of collaborative AI researchers. To escape local minima in the discovery, we propose to encourage orthogonal exploration by constructing the team

Qwen-Audio-3.1-ASR Technical Report

👍 12

10/20 08:00

We present Qwen-Audio-3.1-ASR, a family of audio-language models that builds upon its predecesor Qwen-Audio-3.0-ASR with improved speech recognition and expanded capabilities in speaker-diarized transcription and general audio understanding. The family retains support for 30 languages and 16 Chinese

VerifierBench: Benchmarking the Generation of Executable Verifiers for LLM Agents

👍 4

10/19 08:00

Agents need reliable ways to validate their outputs, both when completing tasks and when filtering synthetic training data. We study verifier generation: from the instruction and environment available for solving a task, an agent writes an executable verifier that must accept every valid solution an

SpatialScout: Can VLM Agents Actively Explore and Map 3D Scenes?

👍 129

10/19 08:00

Spatial intelligence requires agents to actively acquire visual evidence and integrate observations into a coherent representation of the three-dimensional world. Yet spatial question answering offers limited evidence of whether models can construct accurate metric maps through autonomous exploratio

Agent World Model in the Wild: Computer-Use Agent Environment Synthesis

👍 64

10/19 08:00

Agent learning is shifting from learning from human data toward learning through interaction with the world. However, scaling such learning directly in real-world computer systems is challenging and can introduce substantial security risks. We therefore propose Agent World Model in the Wild (WildAWM

Design Anything, Design at Scale

👍 3

10/17 08:00

As multimodal resources are transformed into media outputs, agentic AI is expected to stay aligned with design intent over long horizons. Designers follow briefs specifying goals, constraints, and deliverables, with later work building on earlier decisions. Recent work reuses design experience acros

Language Models Learn to Improve Language Models

👍 5

10/16 08:00

Large language models (LLMs) are increasingly used to automate AI research, including language-model (LM) development, offering a potential path toward recursive self-improvement. However, turning experimental feedback into a learned ability to propose effective improvements across models faces two

Entertain, then Cancel: How Thinking Makes Reasoning Models Read Some and Or Literally

👍 12

10/15 08:00

When a speaker says ``some students passed'', listeners infer that not all of them did. Reasoning models now let users decide how long a model thinks before it answers, but we do not know what extra thinking does to such everyday inferences. We study this directly: we keep a model's weights fixed, c

Spectral-Spatial Interpretation

👍 45

10/15 08:00

Post-hoc image explanations usually show where a model looks using spatial heatmaps or indicate which channels matter using global channel scores. These separate views are limiting for multispectral data because channel relevance is often spatially localized and physically meaningful. We introduce S

PrismaCoder: Pushing the Boundaries of Open Multimodal Code Intelligence through Decomposed Rubric Synthesis

👍 49

10/14 08:00

Multimodal code intelligence connects the symbolic world of programs with the visual world their execution creates. Interest in this capability is burgeoning, as models are increasingly asked to produce charts, webpages, animations, and semi-structured documents. Recently, proprietary models have pu

DeLoop: Repeated Actions as Self-Supervision for Long-Horizon Agents

👍 18

10/14 08:00

Language-model agents that run long tasks often redo work they have already done: they re-issue a tool call they made earlier and get back exactly the same observation, a step that adds no new information. We identify this failure as a loop. Loops appear natively in Qwen, Llama, and Gemini agents an

CompactSplat: Spatially Adaptive Gaussian Distribution for Feedforward Scene Reconstruction

👍 12

10/09 08:00

Reconstructing 3D scenes from sparse views without per-scene optimization remains highly challenging, especially for recovering accurate geometry and fine textures. While recent feedforward approaches leverage generalizable 3D Gaussian Splatting (3DGS) for scene generation, they typically assign one

Right Screen, Wrong Transition: World Models as Verifiers for GUI Agents

👍 81

10/08 08:00

A login screen that appears after a tap on Sign in is expected; the same screen after a tap on View order is an attack. For GUI agents, safety is therefore a property of the transition rather than of the screen, and a monitor that inspects only screens can be defeated by reusing a legitimate one. Ju

How Harvey is Building a Frontier Marketing Org

@gabepereyra · 13.4K 粉丝 · 111.8K 阅 · 505 赞 · 47 转

10/22 01:06

Recently, @figma asked us what Harvey was building with its MCP server. This is the long answer. Between the MCP and better frontier models, agents now run whole design projects end to end across our

中文介绍 分享 Harvey 如何使用 MCP 服务器构建前沿营销团队，涉及 MCP 与前沿模型应用，强调端到端设计项目协作。

How To Use Decision Models for Evals

@HamelHusain · 61.6K 粉丝 · 101.6K 阅 · 1.2K 赞 · 108 转

10/21 00:54

Co-authored with Antaripa Saha. Decision models have exploded onto the scene recently, starting with TypeSafe’s Jev, followed by Perplexity’s Decider, Liquid’s d1, Cloudflare’s Clef, and OpenAI’s

中文介绍 与 Antaripa Saha 合著，介绍决策模型的应用，从 TypeSafe 的 Jev 到 OpenAI 的 Clef，探讨决策模型在评估中的应用。

Using AI to Delete Code: Agentic Tech Debt Cleanup at Uber

@UberEng · 71.4K 粉丝 · 39.8K 阅 · 548 赞 · 53 转

10/21 01:59

Introduction At Uber, feature flags (dynamic Boolean parameters) are central to how we ship software. They let engineers roll out features gradually, run experiments, and roll back quickly if

中文介绍 Uber 工程团队介绍如何利用 AI 删除代码，通过特征标志管理软件发布，实现渐进式功能推出和快速回滚。

The Next Era of Consumer AI: Gotchi Labs' Vision for Everyday Intelligence

@sleepagotchi · 161.6K 粉丝 · 10.0K 阅 · 953 赞 · 72 转

10/20 22:00

From AI that answers questions to AI that understands context, takes action, and powers everyday life. The next generation of consumer technology will not be defined by how intelligently software

中文介绍 探讨消费级 AI 的新时代，Gotchi Labs 展望人工智能在日常生活中的应用，从问答到情景理解与行动。

Designing an AI serving harness built to last

MIT Tech Review AI · 10/23 05:17

The next competitive advantage in enterprise AI has less to do with which model you choose than with what you build around it. That’s an important shift. For the past several years, much of the AI conversation has centered on model capability: bigger models, better reasoning, larger context windows,

中文介绍 企业AI的竞争优势不再仅取决于模型选择，而是围绕其构建的体系，近年来AI对话主要集中在模型能力上。

Build more responsive apps with the Decisions API

OpenAI News · 10/23 03:00

Fast, focused, low-cost decisions for any application.

中文介绍 OpenAI推出快速、专注、低成本的Decisions API，适用于任何应用程序。

The Foundations of Scalable AI

MIT Tech Review AI · 10/23 02:48

A scalable AI system requires a strong foundation. Data, engineering, platforms, operating models, and governance must converge to work as one. This session from MIT Technology Review’s EmTech Future conference explores what it takes to build that foundation in a highly regulated environment and how

中文介绍 可扩展的AI系统需要强大的基础，数据、工程、平台、运营模式和治理必须协同工作。

Why AI Compute Comes in Rectangles — Erik Bernhardsson, Modal CEO

Latent Space · 10/23 01:40

Elastic GPUs, sandbox growth, and the economics of the agent cloud

中文介绍 Modal CEO Erik Bernhardsson讨论了弹性GPU、沙盒增长和代理云的经济效益。

OlmoEarth v1.3: State-of-the-art embeddings for the whole planet

Hugging Face Blog · 10/23 00:11

中文介绍 Hugging Face推出OlmoEarth v1.3，提供全球土地的森林类型嵌入。

Beyond tree cover: mapping the world’s forest types

DeepMind Blog · 10/22 22:30

A global map categorizing every 10-meter patch of land on Earth into distinct forest types.

中文介绍 DeepMind发布全球森林类型地图，将地球每10米土地划分为不同的森林类型。

You are the Team — Anekha Sokhal, Moshi

Latent Space · 10/22 22:03

How Moshi turns market data into actionable insights, and why AI tools have to fit the trader’s workflow.

中文介绍 Moshi将市场数据转化为可操作见解，强调AI工具需适应交易者工作流程。

Making the case for context-aware AI

MIT Tech Review AI · 10/22 18:00

Enterprise AI can generate recommendations, automate tasks, and help organizations make decisions faster. But without an understanding of how a business actually operates, even powerful models can miss the constraints and relationships that determine whether a decision is the right one. As companies

中文介绍 企业AI可生成推荐、自动化任务并快速决策，但需理解业务实际运营才能避免错误决策。

AI’s hidden storage bill: What inference leaves behind

MIT Tech Review AI · 10/22 16:41

Ask most executives to picture the cost of AI and they’ll likely see GPUs—racks of accelerators with gigawatts of power—and a waitlist for silicon. But that picture leaves something out: Data storage costs. What happens to data after a model responds? Much of the back-end data that’s generated—from

中文介绍 AI模型响应后的数据存储成本常被忽视，实际成本可能远超预期。

Ensemble forecasting with IBM Granite time-series models

Hugging Face Blog · 10/22 15:51

中文介绍 Hugging Face介绍IBM Granite时间序列模型，用于集成预测。

[AINews] OpenAI’s Patch the Planet program is 1 of 2 parties to discover PHP2Shell vulnerabilities in core PHP affecting 70% of the web, using GPT-5.6 Sol; coordinated response over 2 months

Latent Space · 10/22 15:42

The Big Hardening begins

中文介绍 OpenAI的Patch the Planet项目发现核心PHP漏洞，影响70%的网站。

What FDE means at OpenAI

OpenAI News · 10/22 14:00

What forward deployed engineering means at OpenAI

中文介绍 OpenAI解释了其前沿部署工程的意义。

Uber brings ChatGPT Work to all employees with OpenAI

OpenAI News · 10/22 08:00

Uber is rolling out ChatGPT Work to more than 30,000 employees, helping teams research, analyze data, create presentations, and complete complex workflows.

中文介绍 Uber通过OpenAI向超过30,000名员工推出ChatGPT Work，帮助团队研究、分析数据、创建演示文稿等。

Claude Duet 🤖, Anthropic API changes 👨‍💻, Recursive’s local model 💻

TLDR AI · 10/22 08:00

中文介绍 Anthropic API更新、Recursive本地模型发布等AI新闻简报。

Broad access through consistent standards

OpenAI News · 10/22 05:00

How OpenAI works with distribution partners to uphold consistent safeguards, support responsible access, and clarify expectations in our Usage Policies.

中文介绍 OpenAI与合作伙伴合作，确保一致的安全措施，支持负责任的使用，并明确使用政策中的期望。