Autoregressive Streaming Architecture Powers Real-Time, Long-Form AI Video with 8X Greater Length Than Industry Standard

July 16, 2025 –  Lightricks, the company behind LTX Video (LTXV) and LTX Studio and a global leader in generative AI video innovation, announced today a major evolution in its AI video technology: an update to LTXV that enables the generation of clips longer than 60 seconds. This milestone positions Lightricks as the first company to enable live-streamed, long-form AI video creation at scale. This breakthrough represents an eightfold leap beyond the current industry norm of eight (8) seconds, and is the first real-time, streaming-capable, long-form AI video model in production today. Unlike traditional models limited to short outputs, LTXV’s autoregressive streaming architecture continuously generates video in real time, enabling developers and creators to build longer, coherent stories - not just isolated clips.

As with prior LTXV updates, this new capability will maintain open weights and be available to developers, academia, and gen AI video enthusiasts. It will also power commercial products like LTX Studio, Lightricks’ flagship creative development platform used by creators and media teams to design, iterate, and produce video projects using AI.

The new LTXV release introduces an improved autoregressive video engine that allows video clips to stream live to viewers as they are rendered. The system returns the first second of content almost instantly and continues generating forward in real time, enabling interactive, continuous narratives with full control over scene development. This paves the way for a new category of generative storytelling applications, from player-generated cutscenes and adaptive educational content to real-time AR visuals synced with live performers.

“Crossing the 60-second mark unlocks a new era for generative media,” said Zeev Farbman, co-founder and CEO of Lightricks. “LTXV is unique in its ability to create longer scenes while maintaining full control of the extended sequences, which enables coherent storytelling with visual and semantic consistency, transforming AI video from a demo or just a random clip, into a true medium with creative intent.”

LTXV’s autoregressive architecture supports both Lightricks' 13B and mobile-friendly 2B parameter models. Creators and developers can apply pose, depth, or canny control LoRAs not only at the beginning of a prompt but continuously throughout the 30-second-plus scene. Compatible with Lightricks' IC-LoRA infrastructure, the system also enables near real-time motion capture feeds, expanding its usability across interactive platforms.

“We’ve reached the point where AI video isn’t just prompted, but truly directed,” added Yaron Inger, co-founder and CTO. “This leap turns AI video into a longform storytelling platform, and not just a visual trick.”

Technical Highlights:

• Real-time autoregressive sequence conditioning across model variants (video generated in chunks of frames, with each chunk conditioning the generation of the next). This allows motion and narrative to be built with smooth continuity, much like a writer creates a story sentence by sentence, each informed by the last.
• Cost and Efficiency: LTXV runs efficiently on a single H100 or even a consumer-grade GPU, while delivering artifact-free 30-second clips. In contrast, public benchmarks for competing solutions show significantly higher compute requirements, with multiple H100s needed to generate just a 5-second 1080p clip (for example, 41 seconds on an L20 and up to eight H100s for higher-resolution real-time output).
• Streaming-first architecture: approximately one (1) second returned instantly, remainder streams live
• Supports continuous control inputs for dynamic scene generation
• Fully compatible with IC-LoRA motion and style LoRA integrations
• Speed - first second returned in ~1 sec, full 60 seconds in real time (builds on the streaming architecture already described)

Key Use Cases Include:

• Advertising and Social Media: 15 to 60-second vertical ad spots generated on demand
• Gaming: Live-rendered cutscenes generated from gameplay data
• Live Events: Stage-synced AR characters reacting in real time
• Education: Adaptive explainer videos that evolve with learner input

LTXV is available as an open weights model on both Hugging Face (LTX-Video) and GitHub (LTX-Video) and already fully integrated into Lightricks’ flagship storytelling platform, LTX Studio. With a growing library of models designed for diverse creative needs and a commitment to open development, Lightricks is shaping the future of generative AI video, bridging research-driven breakthroughs with real-world application. For more information about Lightricks, its products, technology, and open source initiatives, visit www.lightricks.com.

‍

Lightricks, a leader in AI-driven content creation technology, today announced the release of its LTX Video 13-billion-parameter model (LTXV-13B) – which may be the most advanced and efficient AI video generation model to date. This substantial upgrade dramatically increases quality while maintaining LTXV’s unparalleled speed generating AI videos. The 13B model is available within the company's flagship storytelling platform, LTX Studio, shared with the open community and is being integrated across the Lightricks portfolio.​

LTXV-13B introduces “multiscale rendering,” a major technical breakthrough that delivers both speed and quality through a layered process. The model drafts in lower detail first to capture coarse motion using fewer resources. This draft then guides the next stages, where the model progressively adds structure, lighting, and micro-motion (and spending time where it matters most). The result is high-fidelity video built through deliberate, multi-scale generation, with render times that can be more than 30X faster than comparable models – without compromising visual realism. 

The new 13B model represents a significant leap forward in Lightricks' generative AI capabilities, offering creators the ability to produce videos with stunning detail, coherence, and control. It utilizes the latest advancements in academia and the open source community, including unsampling controls and spatiotemporal guidance for video editing, and kernel optimization for running speeds. 

Unlike other models that demand enterprise-grade GPU (and long rendering times), LTXV-13B delivers studio level video at unmatched speed, even on devices that creators already own, which differentiates LTX Video in the marketplace.

"The introduction of our 13B parameter LTX Video model marks a pivotal moment in AI video generation with the ability to generate fast, high-quality videos on consumer GPUs,” said Zeev Farbman, co-founder and CEO of Lightricks. “Our users can now create content with more consistency, better quality, and tighter control. This new version of LTX Video runs on consumer hardware, while staying true to what makes all our products different - speed, creativity, and usability.”

While developing and refining the 13B model, Lightricks entered into a strategic partnership with leading media asset provider Getty Images. In December 2024, Lightricks entered an agreement with Shutterstock to leverage their licensed content. These kinds of collaborations have given Lightricks access to an extensive library of high-quality video assets for model training, reinforcing its mission to build ethically trained, visually compelling, and commercially safe generative tools. 

The LTXV-13B model now empowers creators with even more control and flexibility, seamlessly supporting all of the platform’s advanced creative tools, including:

• Keyframe editing
• Camera motion control
• Character and scene-level motion adjustment
• Multi-shot sequencing and editing

In support of startups and small businesses, Lightricks is offering the 13B model free to license for enterprises with under $10 million in annual revenue. This initiative and the release of all LTXV models in open source reflect Lightricks’ commitment to making cutting-edge generative AI accessible to the next generation of creative companies and innovators. Open source versions of LTXV are available on Hugging Face (LTX-Video) and GitHub (LTX-Video). 

“By consistently refining our models and working with the open community, we’ve built an AI system that generates physically natural movement while preserving artistic control,” added Yoav HaCohen, Director of LTX Video at Lightricks.

Since launching LTX Video in November 2024, Lightricks has collaborated with researchers and open-source contributors to improve motion consistency, scene coherence, and creative adaptability. Key open-source advancements in LTXV-13B include:

• VACE Model Inference – advanced video generation and editing tools, including reference-to-video (R2V). Details on GitHub
• Unsampling Controls for Video Editing – Tools that reverse noise and refine frame granularity. Details on GitHub.
• Kernel Optimization – Efficient Q8 kernel usage allows performance scaling on lower-resource devices. Details on GitHub and HuggingFace.

With a growing library of models designed for diverse creative needs and a commitment to open development, Lightricks is shaping the future of generative AI video, bridging research-driven breakthroughs with real-world application. For more information about Lightricks, its products, technology, and open-source initiatives, visit www.lightricks.com.

‍The integration of FLUX.1 brings unparalleled realism, accuracy, and enhanced workflows that will empower visual storytellers to bring their visions to life with precision and speed 

Lightricks, the company behind AI-driven creative development  platform LTX Studio and the globally successful portrait editor Facetune, announces the integration of FLUX.1, Black Forest Labs’ generative AI image technology, into Lightricks AI-driven filmmaking platform LTX Studio. With this addition, LTX Studio is setting a new standard for creative control, blending cutting-edge technology with intuitive workflows to reimagine how stories are developed.

At the heart of LTX Studio is a unique approach to creative development: creators start with generating high-quality images to visualize their concepts and characters, then seamlessly progress to crafting detailed video sequences—all while maintaining full control over every stage of production. This level of flexibility ensures creators can explore, refine, and iterate with confidence, resulting in projects that align perfectly with their vision.

The integration of FLUX into LTX Studio leverages Lightricks’ proprietary systems, including optimized color spaces enhancing image quality and a refined prompt modeling framework delivering more accurate character consistency. These innovations amplify the performance of FLUX, ensuring creators experience superior results while maintaining industry-leading speed in iterative video generation and creative refinement. Coupled with ongoing updates to LTX Studio’s models and user experience, this release establishes LTX Studio as the ultimate platform for creators seeking precision, quality, and intuitive workflows.

“Flux’s technology takes realism and precision to new heights, perfectly complementing our focus on giving creators unparalleled control over their storytelling process,” said Zeev Farbman, CEO and Co-founder of Lightricks. “The combination of Flux’s revolutionary technology with Lightricks’ intuitive tools empowers creators to visualize ideas and bring them to life faster and more effectively than ever before.”

‍

‍What’s New in LTX Studio?

• True Results & Better Prompt Adherence. New prompt modeling pushes the ability to generate images that closely match complex prompts while maintaining consistency and reliable results to expectations across the entire project, from scenes to characters and styles.

• Realistic, High-Quality Image Generation. A combination of FLUX and enhanced modeling techniques from Lightricks create even more realistic images and scenes, while maintaining consistent character traits including faces, hair, clothing, and body type—even across complex angles and varied environments.

• Improved Scene Consistency Across Projects. Reliable style preservation across locations, lighting, etc., ensures seamless artistic continuity throughout scenes. And locations now retain their defining features across multiple shots, delivering immersive storytelling with consistent environment details.

‍

“We created Black Forest Labs to drive innovation in generative image and video model development and make these technologies as accessible as possible. We are excited to partner with Lightricks to enhance their LTX Studio tool, and continue to explore new frontiers,” said Robin Rombach, CEO and Co-founder of Black Forest Labs. 

Experience Seamless Storytelling Across the Ecosystem

FLUX’s transformative capabilities are also being extended into Lightricks’ popular consumer products, Facetune and Photoleap. Facetune now offers lifelike retouching with unmatched precision, while Photoleap empowers users to create intricate designs with greater realism and consistency.

A New Era of Consistent Creativity

LTX Studio’s FLUX integration introduces “consistent creativity across your entire project,” addressing common challenges in visual storytelling. By combining revolutionary AI technology with advanced development, LTX Studio empowers creators to focus on storytelling while the platform ensures visual continuity, accuracy, and style.

To explore LTX Studio or experience FLUX’s impact across Lightricks products, visit www.lightricks.com or ltx.studio.

‍

‍Shutterstocks' partnership with Lightricks is the first of a new data licensing approach that builds new ways for training AI models 

Lightricks, a global leader of AI-powered creative technology, today announced it is partnering with Shutterstock, Inc. (NYSE: SSTK) to license and use Shutterstock’s extensive video asset library. This collaboration will empower Lightricks to train its open-source video generation model, LTX Video (LTXV), using high-quality HD and 4K video assets. Lightricks is the first global partner to train on Shutterstock’s industry-first "research license," a unique option that reduces the entry barrier for open-source model training. 

Leveraging the research license model mirrors Lightrick’s vision to expand access to creative tools, while driving high quality generative AI results ensure creators’ have protection for their content from the outset. By first integrating a research licence, startups and AI companies can build and refine AI tools on premium, licensed data before making a larger commitment in a full commercial license. This approach also offers a cost-effective path to accessing high-quality data at the research and development stage, addressing a key challenge many new or established companies may face in training AI models.

"This long-term partnership with Shutterstock marks a strategic milestone for us, building on our licensing initiatives from the past year, including agreements with major broadcasters and key content providers." said Zeev Farbman, Co-founder and CEO of Lightricks. "At our core, we’re committed to building powerful, accessible creative tools. This collaboration not only enhances our ability to train AI models with premium, diverse video data, but also elevates the quality and relevance of our generative video tools for our growing community of professional creators and enterprises."

This partnership also continues Lightricks’ AI mission to drive access to creative tools, while driving high-quality generative AI results. By incorporating Shutterstock-licensed video data, Lightricks strengthens its training models, ensuring they are built on robust, industry-recognized datasets that respect the creativity of filmmakers and broadcasters. This announcement follows Lightricks’ release of LTXV 0.9 to the open-source community in November 2024. By leveraging Shutterstock’s premium HD and 4K video data, Lightricks aims to prioritize quality in its next model update, enhancing results for users while advancing its open-source initiatives.

“With our innovative research license model, we empower partners to train their AI models using trusted, licensed data from the start,” said Daniel Mandell, Global Head of Data Licensing & AI at Shutterstock. “We’re thrilled to collaborate with Lightricks on their cutting-edge video tools and remain committed to delivering the data assets they need to achieve long-term success."

Shutterstock’s expansive library—consisting of more than one billion royalty-free assets, including images, videos, music, 3D models, templates, and more—provides one of the largest and most diverse global media datasets available for multimodal model training.

‍About Shutterstock, Inc.

Shutterstock, Inc. (NYSE: SSTK) is a premier partner for transformative brands, digital media and marketing companies, empowering the world to create with confidence. Fueled by millions of creators around the world and a fearless approach to product innovation, Shutterstock is the leading global platform for licensing from the most extensive and diverse collection of high-quality 3D models, videos, music, photographs, vectors and illustrations. From the world's largest content marketplace, to breaking news and A-list entertainment editorial access, to all-in-one content editing platform and studio production service—all using the latest in innovative technology—Shutterstock offers the most comprehensive selection of resources to bring storytelling to life.

‍