Back to Blog
Product Updates

Introducing LTX as an Open World Models Company

LTX announces its evolution into an open world models company, building foundational AI that understands and simulates physical reality for film, robotics, and industry

Key Takeaways
  • LTX is now an open world models company, building AI that understands and simulates physical and virtual environments.
  • Since LTX-2 launched in January 2026, LTX has become the top open model on Hugging Face with 22M+ downloads and production-grade output at a fraction of the cost.
  • LTX models stay fully open for enterprises to run, fine-tune, and extend on their own hardware, under continued leadership from co-founder & CEO Zeev Farbman.

LTX is now an open world models company. We are officially spinning LTX off from Lightricks to become an independent company that provides companies AI tools that enable them to understand, shape, and simulate virtual and physical worlds.

The next wave of AI innovation will play out across filmmaking and gaming to robotics and manufacturing. These are fields where systems need to understand not just how the world looks, but how it behaves. We're building open foundational AI models that understand, shape, and simulate those worlds to empower companies that operate in the real world to be more efficient, and build better, especially in challenging environments.

Over the last few years, LLMs have become the reasoning engine across software. Going forward, world models will become the general-purpose engine for simulating physical reality. They will help us generate environments with an understanding of physics to predict how objects move, how forces interact, and how scenes evolve.

The applications are already broad, and the opportunities will only expand. Animation studios and VFX houses integrate AI into existing production pipelines. Warehouse robots rehearse thousands of runs across a simulated factory floor before operating in a real one. Autonomous vehicles recreate rain-slicked streets at night, running millions of edge cases too dangerous to stage on real roads. In construction, models predict how a building holds up under an earthquake before breaking ground. We even envision world models that let a rover learn a planet's terrain before it launches. The better the world model, the more capable the system built on top of it.

LTX has spent 18 months developing open world models that already understand how light moves, how objects interact, and how scenes evolve. While most of the industry builds closed systems, we believe the future of world models will be open. Enterprises need full control to run, fine-tune, and extend models on their own hardware, within their own environments, and at the edge.

Since releasing LTX-2 in January 2026, we've become the top open world model with 22M+ downloads on Hugging Face. That traction has translated into enterprise demand across video AI and world models, with partners ranging from Fortune 500 companies to VFX houses, production studios, and early-stage robotics startups. LTX outputs are production-grade by design: native 4K, synchronized audio-video, up to 50fps, at roughly one-eighth the cost and one-tenth the render time of comparable models.

LTX will continue to be led by co-founder and CEO Zeev Farbman and his co-founding team. As part of the spin-off, Lightricks will continue to operate and develop its suite of leading consumer mobile apps, including Facetune. Asaf Porat will serve as General Manager of Lightricks.

The shift from AI video to world models is a natural one, and we're building it in the open. We invite developers, studios, and enterprises to build the next era of physical AI alongside us.