Lightricks is raising the bar for AI-generated video with the launch of LTXV-13B, a next-gen model designed to deliver studio-quality results at record-breaking speeds—all on consumer hardware. As rivals like OpenAI’s Sora, Meta, and Google’s Veo 2 continue to dominate headlines, Lightricks is taking a bold, community-driven approach by releasing its most advanced model yet as open source.
Building on the success of its earlier LTX Video release, which stunned creators with its speed and output quality, LTXV-13B packs 13 billion parameters, making it a massive leap from its 2B-parameter predecessor. This upgraded version introduces multiscale rendering for faster, more precise control over visual detail, empowering creators to produce consistent, high-quality videos with ease.
Zeev Farbman, CEO of Lightricks, said the new model remains true to the company’s core principles: speed, creativity, and accessibility. And thanks to its compact architecture, LTXV-13B still runs efficiently on standard consumer devices, even as it delivers professional-grade output.
Open Source, Open Innovation
In a field increasingly locked behind paywalls and proprietary APIs, Lightricks stands apart by keeping its model freely available to the public. With LTXV-13B, the company is doubling down on its commitment to open-source innovation—making the model downloadable via Hugging Face and GitHub with a commercial license exception for businesses earning under $10 million annually.
Farbman emphasized that keeping models open isn’t just about access—it’s about enabling a vibrant, collaborative ecosystem. Individual developers, academics, and hobbyists can now fine-tune LTXV-13B, train it on new datasets, add features, or build their own applications on top of it.
This philosophy runs counter to the locked ecosystems of OpenAI’s Sora or Adobe’s Firefly, which limit experimentation and prohibit model customization. For sectors like gaming, where stylized graphics and rapid prototyping are essential, closed models simply don’t offer the flexibility developers need.
Community Power and Ethical Grounding
The LTXV-13B release benefits significantly from community-driven improvements. Contributors have enhanced motion consistency, scene coherence, and creative adaptability, raising the bar for what’s possible with open-source AI video. Advanced editing features like upsampling controls and VACE model inference now enable workflows such as masked video-to-video, reference-to-video, and style transfer with ease.
Performance remains a priority too. Thanks to Q8 kernel optimization and diffusers, LTXV-13B delivers fast, scalable inference on low-power hardware—critical for democratizing AI video creation.
Perhaps just as important is the model’s ethical foundation. Unlike many competitors, LTXV-13B was trained entirely on licensed assets from Shutterstock and Getty Images. This ensures not only legal clarity but also higher-quality training data—resulting in cleaner, more visually stunning output. As AI companies face mounting scrutiny for unauthorized content scraping, Lightricks’ transparent approach stands out.
Studio-Level Control Without the Studio Price
LTXV-13B is seamlessly integrated into LTX Studio, Lightricks’ web-based editing platform used by creators to produce cinematic content without expensive gear or production teams. The platform offers keyframe editing, multi-shot sequencing, and camera motion control, giving users deep control over their video workflows.
Moreover, LTX Studio lets users tap into external models like Veo 2 and Flux, unlocking even more creative flexibility. While developers can download and customize LTXV-13B locally, many will appreciate the polished, user-friendly experience offered by the integrated Studio suite.
Staying Ahead in the AI Video Arms Race
The release of LTXV-13B comes hot on the heels of LTXV 0.9.6, which already improved rendering resolution, prompt accuracy, and frame rate. In just weeks, Lightricks has followed up with a significantly larger and more powerful model, signaling how quickly the AI video space is evolving.
From Runway’s Gen-4 to Moonvalley’s Marey and Alibaba’s Wan 2.1, the competition in AI video is heating up. But few are balancing performance, openness, and ethics like Lightricks is. With LTXV-13B, the company is positioning itself not just as a tech innovator, but as a champion for accessible, responsible AI.
