AI research papers.

The vision-language-action (VLA) and world-action model (WAM) communities debate whether control policies should inherit pretrained knowledge from vision-language models or from visual generators.

Agentic services alternate between model inference on clustered GPUs and external tools on distributed CPUs.

Language model agents are post-trained and deployed through a harness, the prompts, tools, and control logic that present a task to the model and execute its outputs.

Video captioning aims to preserve visual information through compact descriptions, predominantly using natural language.

We present EmbeddingGemma 2, an open multimodal embedding model which natively represents text, images, audio, video, and combinations thereof in a unified space.

Vision--Language--Action (VLA) models generate robot actions from visual observations and language instructions, but their understanding of physical dynamics remains limited.

Mixture-of-Experts (MoE) models are bottlenecked by loading routed-expert weights from memory during interactive decoding at small batch sizes.

We introduce Timestral, a family of four time series foundation models for probabilistic forecasting, with 5M, 28M, 123M, and 209M parameters.

When an LLM group's discussion is compressed into a short memory for a final decision maker, what should the memory keep?

Visual generation requires both high-level semantic knowledge and fine-grained pixel details.

Scientific discovery often begins with an unexpected observation that prompts a new research question.

Discrete visual tokens represent vision in a discrete form analogous to language, offering a native interface for multimodal modeling under a shared next-token prediction objective.

Vision-Language-Action (VLA) models have made substantial progress in robotic manipulation, but their generalization remains constrained by the diversity and coverage of action-labeled robot demonstrations.

Web agents require diverse, executable, and resettable website environments for reproducible training and evaluation.

OpenAI shares model results on the Kakeya conjecture, Exponential Time Hypothesis, and quantum chromodynamics—and how researchers can verify and build on them.

Explore OpenAI Signals data on 165 work activities across nearly 100 countries, revealing how people use ChatGPT for work and how usage changes over time.

How much of zero-shot benchmark accuracy is captured by word-level alignment between a pre-training corpus and the benchmark target?

Diffusion language models (DLMs) promise parallel text generation, but in practice face an unresolved speed-quality trade-off.

Despite rapid advances in large language models (LLMs), their deployment in the sports and health domain remains high-risk due to the absence of domain-specific evaluation benchmarks.

Language model agents often abandon correct work when a user challenges it, yet simply telling them to stand firm makes them ignore real mistakes.

Vision-language models are shifting from offline passive understanding toward online proactive interaction.

Safe household navigation requires robots to regulate clearance and speed according to what an obstacle is and how close it is, without unnecessary conservatism.

AI agents must decide when another search is worth its cost.

Multi-agent systems (MAS) require both effective coordination and specialized task execution.

Long-horizon agents continuously encounter reusable procedural knowledge, yet existing systems typically rely either on model adaptation or external memory, trading persistent learning for rapid reuse.

An agent that truly knows and acts on behalf of its user has been a long-standing promise of AI assistants.

Reasoning language models generate thousands of tokens per answer, and generation speed is limited by memory bandwidth, so storing weights in two bits instead of bfloat16 can potentially make inference 8times cheaper.

Outcome RL raises pass@1 but not pass@k at large k (Yue et al., 2025).

Physical reasoning underpins how embodied systems understand and act in the physical world, yet vision-language models (VLMs) remain unreliable on questions about dynamics and physical causality.

Long-video generators increasingly retrieve past visual observations to prevent identity drift.

Long-horizon LLM agents interact with their environments through complete assistant turns while generating each action token by token.

Evaluating human-AI co-design requires testing whether coding agents can translate real users' multimodal requests into working application revisions.

AI systems built around large models increasingly use experience and feedback to make persistent changes to models, prompts, memories, tools, and workflows.

Diffusion language models (DLMs) generate text through iterative denoising, enabling bidirectional context modeling and parallel token refinement.

A fundamental challenge for explainable and accountable language models is revealing how external information influences their answers.

Constrained decoding guarantees that Large Language Models (LLMs) produce outputs conforming to predefined formats, but is often assumed to inherently degrade reasoning.

Embodied agents are expected to accomplish real-world goals by deciding for themselves how and when to interact with their surroundings.

Recursive self-improvement (RSI), in which a system also improves the way it improves, is a long-standing aspiration on the path to AGI.

Omnimodal large language models (Omni-LLMs) are increasingly expected to assist users while observing continuous audiovisual streams.

Referring expression grounding in overhead imagery requires oriented bounding boxes rather than horizontal ones, since aerial objects appear at arbitrary orientations.

Critic-free methods are widely used in reinforcement learning with verifiable rewards (RLVR) for long chain-of-thought (CoT) reasoning, avoiding difficult and costly critic training.

As frontier language models grow increasingly capable and autonomous, independent third-party evaluations are critical for characterizing their systemic safety risks.

Generating videos of a specific subject from a few reference images remains challenging when the camera reveals surfaces that are not visible in the input.

High-resolution video matting requires preserving fine details, such as hair strands and soft transparency, while maintaining consistency across frames.

As foundation models evolve from information generation to digital work execution, agents are increasingly capable of completing complex tasks across tools, applications, and workflows.

Image and video generators can produce spatially consistent visual content, but how to use this capability to improve spatial understanding remains underexplored.

Action-conditioned world models offer a scalable alternative to costly real-world interaction for improving robot policies.

Multimodal intelligence can be viewed as human-guided co-evolution between worlds and agents through interaction.

Financial institutions operating under data-residency constraints are restricted to open-weight models small enough to locally host, which trail frontier systems on multi-step quantitative reasoning.

The quality with which language models follow instructions falls as the load grows, yet there is still no quantity that expresses the limit of a model in units of load with an uncertainty.