Multimodal video generation model
A production-grade foundation model deployable via API, on-prem, or open weights.
Licensing
LTX-2.3 is fully open source, free for companies under $10M in revenue. For teams building at scale, we offer a commercial license tailored to your product, infrastructure, and support needs.
API
Production-ready infrastructure for teams shipping AI into real products, with predictable performance, clear pricing, and zero guesswork.
LTX-2.3 Model
Sharper detail. Cleaner audio. Stronger motion. Native portrait. Better video generation.
.svg)
Our Vision
We believe creative systems should be reliable enough for real production, flexible enough to adapt to different workflows, and open enough to improve through use. This is why we’re building an open, production-grade system for cinematic, synchronized audio–video creation.
Innovation
Research
Research
LTX-2.3 is built on a 22B-parameter asymmetric dual-stream diffusion transformer: a 14B video stream and 5B audio stream with bidirectional cross-attention. The full paper, weights, and code are publicly available.
Asymmetric dual-stream DiT architecture. Joint audio-video generation with bidirectional cross-attention and modality-aware classifier-free guidance. Open weights and code.
Rebuilt VAE for sharper detail. 4x larger text connector for tighter prompt adherence. Native portrait, cleaner audio, HDR output, and precise control over motion and camera.
For academic teams pushing the boundaries of video generation, world simulation, and multimodal AI. Grants, model access, and research partnerships with the LTX team.
WHAT'S NEW
Production-grade reframing.
Outpaint video beyond its original frame while preserving composition, motion, visual consistency, and on-screen elements.
Designed for production workflows where consistency matters.
Open Source
Open source, by design
An open source video generation model built to evolve in the real world. Excellence doesn’t happen in isolation. It emerges through real use, iteration, and collaboration. Open source is how we raise the bar — together.
Model weights, code, and core tooling are openly available for inception, extension, and reuse.
Improvement driven through real-world usage, iterative experimentation, and community driven collaboration.
Built on LTX
LTX Desktop.
Running on your machine.
A full video editor built on the LTX-2.3 engine, running locally on your hardware. Open weights, no cloud dependency, released as open software. Run it locally or integrate the engine via API
The LTX Stack
Build, create, and scale with LTX
Production-grade video generation models designed to hold up under real workloads. Built for long sequences, precise motion, and high-fidelity output from fast iteration to final-quality renders. Learn More →
HDR Output
Delivered as an IC-LoRA on LTX-2.3. Generate directly in HDR or convert existing SDR footage to EXR. More grading latitude, more range, ready for real finishing pipelines.
Native Portrait
Generate vertical video up to 1080×1920 — trained on portrait-orientation data, not cropped from landscape.
Audio to Video
Generate video where voice, music, and sound effects define structure, pacing, and motion.Built for production-grade workflows that require precise, harmonious control over audio-led scenes - from podcasts and avatars to voice-driven clips -not one-off demos or talking heads.
20 sec Clip
Extend creative range with long-form generation. Produce up to 20 seconds of high-fidelity video with complete control and consistent style.
Customer Voices
Success, engineered together
Training and steering video models like LTX is the most viable way to align AI with real production needs, where predictability, ownership, and creative intent matter as much as visual quality"
FAQs
What are LTX Models?
LTX Models are multimodal video generation models designed for developers and product teams. They combine text, image, and video inputs to generate, animate, and edit video through a scalable API, enabling workflows like text-to-video, image-to-video, and selective video editing.
What can I build with LTX Models?
LTX Models can be used to build production-grade video systems, not just demos.
Common use cases include:
- Text-to-video and image-to-video generation
- Automated video editing, retakes, and refinement
- Creative tools and content platforms
- Internal production pipelines for studios, brands, and agencies
- AI-powered video features embedded inside larger products
LTX is designed to integrate directly into existing workflows — via API, open weights, or custom deployments.
Who are LTX Models built for?
LTX Models are built for teams that need reliable, scalable video generation, including:
- Developers and startups building AI-powered video products
- Creative platforms and AI hubs aggregating multiple models
- Studios, agencies, and in-house production teams
- Enterprises integrating video generation into real production systems
- Academic and research groups working with open, inspectable models
If you need predictability, control, and the ability to ship — LTX is designed for you.
Can I integrate LTX Model into my own product or platform?
Yes.
Many teams use LTX Model as a core capability inside their own products — either as a primary feature or as an embedded service.
You can integrate LTX Model via:
- The LTX API (managed, usage-based)
- Open model weights (self-hosted or customized)
This makes LTX Model suitable for both startups and large platforms that need flexibility around deployment, pricing, and control.
Is LTX Model suitable for real production use?
LTX Model is designed for production workloads, with a focus on:
- Stable, predictable outputs
- Long-form video generation
- Consistent quality across runs
- Performance that scales from small teams to large deployments
This is not a research demo or experimental-only release.
How does open source fit into LTX Models?
LTX is open by design.
Model weights, core code, and tooling are available for inspection, extension, and reuse.
Teams can:
- Run LTX locally or on their own infrastructure
- Fine-tune and customize models for specific domains
- Integrate deeply without vendor lock-in
For many organizations, this openness is critical for trust, experimentation, and long-term ownership.
Can LTX Model be deployed on-prem or in isolated environments?
Yes.
LTX Model supports self-hosted and isolated deployments for organizations with strict requirements around:
- Data privacy
- Security
- Regulatory compliance
- IP protection
This makes it suitable for enterprise, financial, governmental, and other sensitive environments.
Is LTX Model only for video generation?
Video is the core focus today, but LTX Model is built as a multimodal system.
It supports workflows that combine text, images, audio, and video, and is designed to expand as multimodal capabilities evolve.
Why is LTX Model is the best video model?
LTX Model is built around three principles:
- Production readiness — designed to work reliably, not just impress
- Openness and control — inspectable, modifiable, and deployable
- Integration-first design — meant to live inside real products and workflows
For many teams, this combination is what makes LTX Model viable beyond experimentation.
Where should I start?
- If you want a managed experience: start with the LTX API
- If you want full control: explore the open model weights
- If you’re evaluating for production: review the documentation and examples
Is the LTX Model free to try?
Yes. You can try LTX for free using the LTX Playground, where you can explore the models, generate videos, and experiment with different inputs before integrating the API into your product.
What is the latest LTX model?
LTX-2.3 is our latest model — delivering sharper output, native portrait video, improved prompt adherence, and cleaner audio. Learn more about LTX-2.3
