LTX Desktop

LTX Desktop is an open-source AI video production suite that combines a full non-linear editor with on-device AI generation, powered by the LTX-2.3 engine. It targets video editors, content creators, and developers who need to generate and edit video entirely on their own hardware. The core value proposition is complete independence from cloud services: no internet connection is required after initial setup, no per-generation fees are charged, and all footage, prompts, and outputs stay on the user's machine. This suite runs locally on Windows with NVIDIA GPUs or on macOS via API, offering a versatile solution for professionals who prioritize privacy, cost control, and operational autonomy. The application is fully open source under Apache 2.0, allowing the community to fork, modify, and build upon it freely.

Traditional AI video generation typically depends on cloud APIs that accrue recurring costs, introduce latency, and force data to leave the local environment. LTX Desktop eliminates these pain points by performing all inference on the user's own hardware. For professionals handling confidential client footage, sensitive corporate content, or projects with strict data residency requirements, this local-only approach is critical. It also removes the unpredictability of API pricing and the need for a constant internet connection. Users can generate and edit video in remote locations or secure facilities without worrying about data egress or service outages. The result is a production workflow that is both cost-predictable and privacy-compliant from the ground up.

The Gen Space within LTX Desktop provides a suite of AI generation capabilities including text-to-video, image-to-video, audio-to-video, still image generation via Z-Image Turbo, and a retake feature that regenerates specific portions of a video. These tools work directly on the user's local GPU, enabling rapid experimentation without any usage fees. For example, a creator can type a text prompt, generate a video clip, refine it, and regenerate only a problematic segment using retake—all without leaving the application. Z-Image Turbo adds the ability to create still images locally, which can then be used as inputs for video generation. This feature set allows users to iterate freely, producing multiple variations of a scene to find the perfect take without worrying about per-render costs.

The integrated non-linear editor (NLE) offers professional trimming tools such as slip, slide, roll, and ripple edits, along with built-in transitions and primary color correction. AI-native editing features include the ability to generate multiple takes per clip directly in the timeline and switch between them non-destructively, with each version nested inside the clip for a modular workflow. Context-aware gap fill automatically generates content that matches surrounding clips, using an optional Gemini API connection for intelligent scene continuation. The editor also supports customizable keyboard shortcuts with presets for Adobe Premiere Pro, DaVinci Resolve, and Avid, making it familiar for editors switching from traditional NLEs. These features blend AI generation seamlessly into the editing process, allowing creators to enhance their timelines without manual rendering or external tools.