AI research papers

arXiv:2609.15322v2 Announce Type: replace-cross Abstract: Pretrained driving vision-language models (VLMs) integrate visual, route, language, and driving context into rich driving priors, yet their representation objecti…

arXiv:2609.16498v2 Announce Type: replace-cross Abstract: Research data repositories are essential infrastructure for scientific inquiry and for ensuring that datasets follow FAIR (Findable, Accessible, Interoperable, an…

arXiv:2609.16737v2 Announce Type: replace-cross Abstract: Generative video models can serve as a promising backbone for robot navigation by predicting future observations as video plans.

arXiv:2609.17171v2 Announce Type: replace-cross Abstract: The interpretability of complex machine learning models is of paramount importance, especially in real-world high-stakes domains such as healthcare and finance.

arXiv:2609.17274v2 Announce Type: replace-cross Abstract: AI agents increasingly act through agent skills, i.e., natural-language instructions, that direct a host agent toward shell, network, credential, file, and proces…

arXiv:2609.17327v2 Announce Type: replace-cross Abstract: This paper describes our submission to the SHROOM-Visions shared task on detecting and classifying hallucinated character spans in vision-language model outputs a…

OpenAI shares a framework for tracking, investigating, and disclosing model misalignment, alongside six reports of unexpected or concerning model behavior.

Multiple cameras tracked a mouse as it wandered around a small arena.

The AI boom is becoming a materials challenge.

This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology.

Current Mixture-of-Agents (MoA) paradigms generally treat query routing and agent fine-tuning as separate processes, limiting their ability to respond to evolving agent capabilities.

Enabling robots to adapt to unfamiliar environments as readily as humans remains a moonshot goal of embodied AI.

Mixture-of-experts (MoE) inference on consumer hardware is bounded by weight memory: a 35B-class model is 19.5GB at 4-bit, and sparsity shrinks the compute per token, not the bytes that must be held.

Intelligent systems that act in the world require image understanding that is both comprehensive and spatially grounded.

In reinforcement learning for large language models, Proximal Policy Optimization (PPO) commonly uses a critic to estimate state values and reduce the variance of policy updates.

Action tokenizers play a central role in autoregressive vision-language-action (VLA) models, determining both the targets for policy training and the executable commands recovered from predicted tokens.

Direct preference alignment methods are widely used to align large language models (LLMs) with human preferences because of their computational and memory efficiency.

Autonomous research loops such as AutoResearch show that one coding agent can improve a training setup unattended.

Scientific code repositories encode decades of human knowledge in executable models, methods, and tools.

In collaborative tasks with asymmetric information, participants coordinate their understanding through interaction.

Coding agents are typically evaluated with desired behavior specified through issues or instructions.

Listen to the session or watch below Employees at the world’s leading AI labs are saying there’s a real possibility that advanced AI could destroy humanity.

This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology.

Last year Ruxandra Teslo, a policy analyst who focuses on clinical trials, posted an idea for supercharging medical AI systems: Use data from failed biotech companies.

When Jessica Wachter, a finance professor at the University of Pennsylvania’s Wharton School, wanted to assess AI’s impact on the economy over the next few years, she faced a long list of business and technical uncertain…

A walking robotic hand must use the same fingers to move its body, support its weight, and interact with the environment.

When agentic sessions run to a million tokens with many sessions resident at once, the KV cache and the index that ranks it live in host memory, and the scan that ranks all n keys for a top-k step becomes the traffic tha…

We introduce LimiX-2, a new model in the LimiX family, developed through model and data scaling guided by our previously established scaling laws.

Reliable confidence estimation is increasingly central to the trustworthy deployment of language models: a calibrated estimate of the probability that an output is correct decides what to ship, what to escalate, and what…

We introduce Zing-0.5, a 5B autoregressive world model designed for playability: users can explore generated worlds, influence unfolding events, and respond to the resulting feedback through joint keyboard and online tex…

Large language model (LLM) trading agents can combine market data, news, and executable analysis, but their behavior is often controlled by static hand-written tool-use policies that are fixed before deployment.

3D hand mesh reconstruction is a challenging yet essential task for downstream applications, including human-robot interaction and AR/VR.

Traditional multimodal representation learning and generation are two stages: a contrastive or self-supervised visual encoder is trained first, followed by a separate downstream generative model.

Language model-generated rubrics are increasingly used as reward signals for rubric-based reinforcement learning, LLM-as-a-judge evaluation, and automated grading.

This story appeared in The Algorithm, our weekly newsletter on AI.

BraTS-GoAT evaluates tumor segmentation across heterogeneous populations.

Scientific agents contribute to hypothesis discovery by synthesizing evidence, assessing proposals, and developing new explanations.

Diffusion Transformers deliver state-of-the-art video generation, but their long spatiotemporal sequences make attention the dominant deployment cost, and a deployable low-bit kernel must be accurate and fast.

Masked diffusion language models (dLLMs) generate text by iteratively denoising masked tokens with bidirectional attention.

Training a Mixture-of-Experts (MoE) model at long context or large batch size fails as soon as any one component's peak allocation exceeds device memory, so the target is every peak at once, not the average footprint.

Recently, linear attention layers have been increasingly adopted to replace softmax attention at scale for long-context modeling.

Unified multimodal large language models (MLLMs) and multi-agent systems have advanced visual generation.

Few-shot in-context learning (ICL), the capacity of a model to infer abstract patterns from input-output examples provided in its prompt and apply them to new inputs, has been extensively studied in large language models…

Open-vocabulary detection accepts any class list at inference, and promptable segmentation returns regions without class names: the taxonomy has left the model and become an input.

When we speak of recursive self-improvement (RSI), are we speaking of a phenomenon, a mechanism, or a prospect?

We’re sharing an AI-generated solution to the Navier–Stokes Millennium Prize Problem, including a writeup and a formal proof in Lean.

Inside OpenAI, coding agents are reshaping AI research.