Providers

World Labs

Spatial intelligence world models from World Labs.

Google DeepMind

Interactive world generation research from Google DeepMind.

Odyssey

AI lab building real-time world models — including Odyssey-2, Starchild-1, and the multi-agent Agora-1 — for interactive simulation, positioning world models as a foundation for gaming, robotics, and beyond.

Decart

Real-time generative AI lab behind Oasis and the Lucy model family, building world models and low-latency infrastructure for physical AI, video transformation, and interactive simulation.

NVIDIA

Developer of Cosmos, an open world foundation model platform for physical AI — generating and predicting physics-based environments to train robots, autonomous vehicles, and simulation systems.

Tencent Hunyuan

Tencent's Hunyuan AI model family's 3D/world-model product line — generates immersive, explorable, interactive 3D worlds from text or images (HunyuanWorld / HY-World).

Ophilus

A frontier lab focused on data foundations for multi-agent interaction. Publisher of Khora, a scalable multiplayer world model built in collaboration with RhOS.ai.

Overworld

Developer of Waypoint-1.5, an open-weight real-time video world model for consumer hardware, and the Overworld Biome runtime / overworld.stream browser platform.

RhOS

Builds physical large models for embodied intelligence across robots, humans, objects, and scenes. Co-developer of Khora alongside Ophilus.

Veeda AI

Toronto-based world model startup building multimodal foundation world models for physical AI/embodied intelligence (robotics simulation/training). Founded by Sanja Fidler (CEO, former head of NVIDIA's Toronto AI lab), Zan Gojcic (CTO, ex-NVIDIA Zurich), and Huan Ling (chief scientist, ex-NVIDIA). Raised $90M+ seed (Khosla Ventures, Radical Ventures, August 2026). Currently in stealth mode — no named or publicly released Model exists yet.

Skywork AI

AI research org behind the Matrix-Game series of interactive world models.

San Francisco-based developer platform for real-time generative AI video and world models — GPU provisioning, session orchestration, and sub-50ms-latency streaming infrastructure for interactive AI applications. Founded 2025 by Alberto Taiuti and Bryce Schmidtchen; raised a $59M Series A in May 2026 (led by Lightspeed Venture Partners). Sponsors and provides the runtime infrastructure behind OpenDreamer, an independent research project built by Diego Marti Monso, Francesco Sacco and Edward Hu — Reactor supported and hosts the work but did not itself author the model.

MBZUAI

Mohamed bin Zayed University of Artificial Intelligence — built the PAN world model.

General Intuition

New York-based frontier research lab building foundation/world models for embodied AI, training on large-scale video game data (via co-founder Pim de Witte's Medal platform, ~2 billion videos/year). Raised $320M at a $2.3B valuation (Khosla Ventures, General Catalyst, Jeff Bezos, Eric Schmidt).

Lightricks

AI technology company behind the LTX model family (LTX-2, LTX-2.5, LTX Studio) and consumer creative apps (Facetune, Photoleap, Videoleap). Describes its mission as building an "Open Creativity Stack" — an ecosystem of AI tools and models for creators.

Alaya Lab

AI x Gaming research lab, founded 2026 — a group of game enthusiasts, AI researchers and creators. Also affiliated with Shanghai Innovation Institute and Huazhong University of Science and Technology on published research.