TLDR: Lightricks has launched LTXV, an open-source AI model capable of generating continuous video content exceeding 60 seconds, a significant leap beyond current industry standards. LTXV features real-time streaming and refinement capabilities, allowing creators unprecedented control and efficiency in producing long-form narratives, and is accessible to individuals and small teams.
Lightricks, the innovative Israeli tech company renowned for applications like Facetune and Videoleap, has announced a groundbreaking advancement in artificial intelligence with the launch of its new model, LTXV. This open-source AI model is set to redefine generative media by enabling the creation of continuous video content that can exceed 60 seconds, a duration significantly longer than what is currently achievable with leading models such as OpenAI’s Sora, Google’s Veo, or Runway’s Gen-4, which typically face an 8-second limitation.
According to Zeev Farbman, co-founder and CEO of Lightricks, LTXV “unlocks a new era for generative media.” The model is engineered with an autoregressive streaming architecture, designed to begin streaming results almost instantly, generating the first second of video immediately and constructing the subsequent sequence on the fly. This innovative approach ensures seamless continuity, maintaining consistent characters, motion, and storyline across extended durations by utilizing overlapping frame chunks.
LTXV distinguishes itself not only by its extended video length but also by its remarkable speed. It reportedly delivers output much faster than competitors like Veo 3, Runway Gen-4, or Kuaishou’s Kling, which often require users to wait several minutes for just a few seconds of video. Furthermore, the model introduces dynamic scene control, empowering users to adjust elements such as posture, depth, or style in real-time during the video generation process. It supports various generation modes, including text-to-video, image-to-video, and keyframe animation.
A key aspect of LTXV’s release is its commitment to open development. The model is open-source and unrestricted by a proprietary API, with its open weights available on platforms like GitHub and Hugging Face. This accessibility aligns with Lightricks’ philosophy of “open development for real-world application,” providing developers and independent artists the freedom to build upon the core engine. It is free to use for individuals and small teams with annual earnings under $10 million. From a technical standpoint, LTXV is remarkably lightweight, capable of running on a single Nvidia H100 GPU or even high-end consumer GPUs, a stark contrast to many proprietary models that demand substantial cloud-based computing resources.
Lightricks co-founder and Chief Technology Officer Yaron Inger emphasized the shift LTXV represents, stating, “We’ve reached a point where AI video isn’t just prompted, it’s truly directed. This leap turns AI video into a long-form storytelling platform, and not just a visual trick.” The company also highlights that LTXV was trained on fully-licensed data from stock image providers such as Getty Images and Shutterstock, ensuring that generated videos are free from copyright infringements.
Also Read:
- OpenAI Partnership Powers Invideo AI: Revolutionizing Video Creation for All
- Adobe Firefly Unveils Major AI-Powered Video and Audio Enhancements
The new capabilities of LTXV open doors to a myriad of new use cases, including professional film production, real-time content generation, adaptive educational and advertising content, augmented reality visuals synchronized with live performances, and dynamic player-driven game cutscenes. This breakthrough positions Lightricks as a trailblazer in the evolving landscape of AI-driven content creation, transforming AI video from short, isolated clips into a true medium for coherent and directed storytelling.
