Lightricks, the creative technology company behind popular visual editing tools, has announced the release of LTX-2, an open-source foundation model that promises to redefine AI-driven video generation. The new model delivers 4K resolution, synchronized sound, and real-time rendering speeds, setting a new benchmark for rapid video production.
Lightricks
According to the company, LTX-2 can generate a six-second, high-definition video in just five seconds—faster than real-time playback—without sacrificing quality. For creators seeking top-tier output, the model can produce 4K footage at up to 48 frames per second, complete with native audio, all while running on consumer-grade GPUs, drastically cutting compute costs.
A Leap Forward for Diffusion Models
LTX-2 operates as a diffusion model, a type of AI that refines generated visuals by progressively removing “noise” from initial data until it reaches a polished final output. What sets LTX-2 apart is its accelerated diffusion process, allowing creators to preview and iterate on videos almost instantly. The system also generates synchronized soundtracks, dialogue, and ambient effects, eliminating the need for separate audio post-production.
Lightricks’ co-founder and CEO Zeev Farbman described the release as a major milestone:
“LTX-2 is the most complete and comprehensive creative AI engine we’ve ever built, combining synchronized audio and video, 4K fidelity, flexible workflows, and radical efficiency,” he said. “This isn’t vaporware—it’s a real breakthrough in video generation.”
Built for Creators and Professionals
LTX-2 will be available through Lightricks’ LTX Studio platform and via API access. Professionals using LTX Studio can opt for slightly slower processing to achieve cinematic-grade output, while developers can experiment freely with the open-source version, slated for release in November.
Lightricks said the model’s upcoming creative controls will include pose and depth editing, video-to-video generation, and rendering customization, enabling users to tailor the system to specific artistic needs.
Continuing a Track Record of Innovation
The launch follows a series of advances in Lightricks’ LTX family of models. Earlier in 2024, the company’s LTXV-13B and LTXV-2B models introduced support for long-form, directed video generation of up to 60 seconds, with real-time prompting. The models also pioneered multi-scale rendering, allowing creators to add depth and color in stages, much like traditional animation workflows.
Charlie Fink
Importantly, Lightricks trains its models using licensed datasets from Getty Images and Shutterstock, helping avoid copyright pitfalls that have dogged other AI video tools.
Patrecia Meliana
Transparent and Affordable Access
Unlike many competitors, Lightricks says LTX-2 is fully open-source, with pre-trained weights, datasets, and tools to be published on GitHub. For those preferring paid API access, the company offers tiered pricing starting at $0.04 per second for HD video generation, $0.07 for Pro performance, and $0.12 for 4K Ultra quality with full audio.
Thanks to its processing efficiency, Lightricks estimates LTX-2 to be up to 50% cheaper than comparable models, offering creators faster iteration and higher-quality results at lower costs.
A Defining Moment for AI Video Creation
With the release of LTX-2, Lightricks continues to position itself at the forefront of AI-driven creative technology. The company’s commitment to open-source accessibility and professional-grade performance could make high-quality video generation more attainable—and more ethical—than ever before.
