Lightricks, the company behind Facetune and LTX Studio, and a pioneer in AI-driven content creation, today introduced LTX-2, its next-generation open-source foundation model.
The company said it is a major leap in creative AI that delivers professional-grade video and sound generation in one system, built for real-world production.
LTX-2 combines native 4K resolution, 50 fps motion, synchronized audio, and 15-second generation lengths with radical efficiency and open-source transparency. It runs efficiently on consumer hardware, delivers outputs at up to 50% lower cost than competing models, and integrates seamlessly into production workflows via the LTX-2 API.
“Diffusion models are reaching a point where they no longer just simulate production – they are production. LTX-2 represents that shift: the most complete creative engine we’ve built,” said Zeev Farbman, CEO of Lightricks, in a statement. “It unites synchronized audio and video, 4K fidelity, long-form capability, and radical efficiency in one open, production-ready system – built to empower everyone, from independent creators to enterprise teams.”
The company’s mission is to bridge the gap between imagination and creation, building world-class AI models, the infrastructure to deploy them, and the applications that bring them to millions of creators and businesses worldwide.
What’s new in LTX-2: Six advances one model
Visuals and sound are created together in a single process – aligning motion, dialogue, ambience, and music with natural timing. This unlocks realistic storytelling across branded content, film, and social formats.
Native 4K Fidelity, 15 Seconds, 50 fps: True 4K resolution at up to 50 frames per second with synchronized audio, supporting clips up to 15 seconds long – bridging cinematic fidelity and creative flexibility.
Efficient, Scalable Performance: LTX-2 delivers professional-grade results with up to 50% lower compute cost than competing models. Its multi-GPU inference stack enables long-form generation with precision and scale.
Consumer Hardware, Professional Output: The model runs efficiently on consumer-grade GPUs, democratizing access to world-class AI video creation. From indie developers to large studios, anyone can now produce professional output without enterprise infrastructure.
Creative Control and Precision: Multi-keyframe conditioning, 3D camera logic, and LoRA fine-tuning provide frame-level control and stylistic consistency. LTX-2 supports multimodal inputs – text, image, audio, depth maps, and reference video – for detailed guidance.
Open source, built for collaboration
Core components of LTX-2 are available on GitHub, including datasets and inference tooling. Model weights will be released later this fall, enabling researchers, developers, and studios to customize, extend, and innovate freely.
Three Flows for Every Creative Need
- Fast (starting at $0.04/sec) – Optimized for responsiveness and iteration. Ideal for previews, ideation, and mobile workflows.
- Pro (starting at $0.08/sec) – Balanced for efficiency and polish. The default for daily production work and marketing teams.
- Ultra (starting at $0.16/sec) – Maximum fidelity for cinematic, branded, and VFX production, delivering 4K video at up to 50 fps with synchronized audio and 15-second sequences.
Pricing varies by resolution (720p–4K) and whether the generation includes synchronized audio.
LTX-2 API and ecosystem availability
API access can be requested through the LTX-2 website and is being rolled out gradually to early partners and teams, with integrations available through Fal, Replicate, ComfyUI, and more.
LTX-2 is also available via the LTX platform, with upcoming marketplace availability later this year. Full model weights and advanced tooling will be released to the open-source community later this year, supporting continued innovation and ecosystem growth.