🎉 OnlyAct is accepted at #NeurIPS2026#! Take the motion from one video, play it in any world 🎬 We turn a pretrained video encoder into an action encoder with zero action labels. 🌐 Page: 💻 Code: #WorldModels# #Robotics# 🧵👇

Coding drives innovation & fuels creativity. Yet the lack of diversity in programming can perpetuate biases & deepen inequalities in the digital world. On Thursday’s #WorldCodingDay#, @UNESCO is calling for a more inclusive digital future.

GPT-5.6 Sol sets a new state of the art in cybersecurity on “The Last Ones” cyber range. We’re already seeing that capability translate into defensive outcomes: helping teams find, validate, and fix vulnerabilities in real-world code. Put it to work with Codex Security:

TL;DR: A survey that treats video as an incomplete projection of a hidden world and organizes video AI research into four operations: Reading, Writing, Sharing, and Interacting. Title: Reading, Writing, Sharing, and Interacting with the World through Video URL: Key points ・👁️ Reading: infer semantics, space, time, physics, and causality from video ・🎬 Writing: generate video from text, images, camera paths, and other conditions ・🔗 Sharing: unify understanding and generation via shared tokens and objectives ・🕹️ Interacting: closed-loop world models that predict state transitions under actions ・🎞️ Temporal coherence (neighboring frames agree) is not world consistency (compatible with an admissible world trajectory) ・🪜 An evidence ladder grades claims from plausible output up to decision utility The shift in emphasis from visual fidelity to persistent state and physical and causal validity is a useful way to read the field. #WorldModels# #VideoGeneration#

🚀 Introducing Hy-Nomas: An Open Source Free-Form Interactive Video World Model Key Features: * Free-form interaction with text prompts and camera trajectories * Long-horizon generation with world memory * Lightweight LoRA adaptation for specialized domains such as robotics * Optimized for Inference on Consumer GPUs Test it and drop feedback anytime！ 🔗 Website: 💻 GitHub: 🤗 Hugging Face: 🎥 Demo: #Hy##WorldModels# #VideoGeneration# #GenerativeAI#

🌍 TL;DR: A training-free way to re-explore a recorded video from new camera angles just dropped. It handles viewpoint changes, hole-filling, and recovering earlier appearances entirely through a frozen model's own attention. Title: World in World: Explore the World with World Models URL: Points 🎥 Four evidence sources (source video, target-view scene, rendered geometry, retrieved history) get converted into clean K/V fed straight into the frozen model 🧭 CGAR routes queries toward reliable source-video tokens using persistent point correspondence and camera geometry 🎚 EWA independently tunes each auxiliary channel's contribution using existing attention outputs, with zero extra network evaluations 🏆 Top VBench score of 85.192 and lowest rotation error (2.8326°) against 6 baselines including ReCamMaster 🔧 Same framework also powers bullet-time rendering, video stabilization, editing, and motion transfer Getting this much versatility with zero retraining feels like a win for evidence engineering over model adaptation. #WorldModels# #VideoGeneration#

A new approach to physical world models — reason about "how the world changes" in language before generating any pixels. Title: PhiZero: A World Model Built Around Physical Language ❓ What is "Physical Language"? 💡 A discrete token sequence that directly represents state-transition patterns in a video — separate from visual appearance. Because dynamics and appearance are disentangled, the state transitions encoded from one video can be transferred to a completely different visual style. PhiZero's "Reason-then-render" paradigm predicts this physical language first, then generates video from it. ❓ Why did previous physical world models produce physically inconsistent results? 💡 They relied on direct pixel-space prediction, burying dynamics implicitly inside the model. Reconciling visual realism with physical consistency was difficult — leading to failures like a tennis ball hitting a rubber duck that doesn't react. The reasoning process was also opaque by design. ❓ What performance numbers did it achieve? 💡 State-of-the-art results across all six benchmarks: ・Physics-IQ IQ-Score: 41.2 (beats previous SOTA of 39.5) ・WorldModelBench: highest in physical compliance, commonsense, and overall score ・YoCausal causal reasoning: aggregated rank 2.0 (best) ・Tokenizer: just 256 tokens achieve PSNR 28.9 — 1/175th of Wan2.2-5B VAE's token count ❓ What does this mean for robotics and autonomous driving? 💡 Physical Language is appearance-agnostic, so zero-shot sim-to-real and human-to-robot transfer become possible. The team demonstrated encoding LIBERO simulation dynamics and replaying them as realistic video, and directly transferring human motion patterns to robot embodiments — all without task-specific training. #WorldModel# #EmbodiedAI#

Every time I fly to New York, I'm struck by how many trees there are in the densely-populated areas around the city. (Coming from the West Coast, one arrives over the Garden State, and the suitability of the sobriquet is quite apparent from the air.) It feels like this degree of tree cover in highly populated areas is atypical (the environs of places like Paris, London, Los Angeles, Melbourne, Tokyo, etc., look very different), so I asked the LLM to investigate. It found GHSL 2020 population and ESA WorldCover 2021 10m land-cover data, and concluded that New York is in fact quite unusual. As far as I can tell, it's because of some combination of: * A very favorable climate. (Trees grow quickly without irrigation.) * Marginal farmland. (Readily outcompeted by the Midwest in the 19th century.) * Together yielding reforestation before the advent of suburbs. * And a preference for development patterns that include trees. (Japan's climate is very hospitable, but one sees far fewer trees in the populated areas around the major cities -- forest and habitation are more disjoint.)

Seeing InternVideo-Next explicitly characterize its architecture as "Encoder–Predictor–Decoder (EPD)" and defining the Predictor as a "latent world model" confirms a clear convergence in the field. 📉 From our early work on Context AutoEncoder (CAE, — where we decoupled the Encoder, Predictor ("Regressor" in the paper), and Decoder — to Meta's I-JEPA/V-JEPA shifting entirely to latent prediction, and now this. It seems we are all validating the same core intuition: Understanding ≠ Pixel Reconstruction. 💡 Decoupling representation learning from high-frequency detail generation to build World Models in latent space is evidently the path forward. Glad to see our early intuition resonating with the latest SOTA. 🫡 #AI# #ComputerVision# #WorldModel# #JEPA# #CAE#

Is a good-looking generated video a world model? WorldFoundry: Towards Unified Infrastructure for World Intelligence A paper that gives a functional definition of world models and an open-source infrastructure to run and evaluate very different models and benchmarks in the same way. Highlights 🎯 Closure, not looks, is the dividing line An intervention must leave a trace in the state that constrains later predictions, and new observations must revise that state. The authors argue this operational persistence is what separates foundation generative models from world models. 🧩 Three operations to organize systems State inference, forward simulation, and interactive response describe any world-model-like system. No system has to implement all three; it is an interface for comparing video, 3D/4D, and embodied models on one scale. 🔧 Evidence-aware scorecards Declared support, operational readiness, and validated performance are recorded separately. It integrates 270 models and 74 benchmarks (the abstract's counts), but validation confirms that things run, not which model is better. Unglamorous evaluation infrastructure could matter a lot for world-model research. #WorldModels# #OpenSource#