Radar / AI media tools / LTX-2.5
LTX-2.5: Lightricks' open-weights video world model with sound
A 22B open-weights audio-video foundation model that generates synchronized video and audio from text, image or video, with native multishot generation that holds character, environment and voice across cuts.
Why it matters
The most capable open video model for filmmakers: multishot scenes in one pass, diffusion-fidelity rendering, and day-one ComfyUI support - free for individuals and companies under $10M revenue.
What you could build with it
An indie film team could use LTX-2.5 in ComfyUI to previz entire sequences - multishot scenes with consistent characters and synchronized sound - before spending anything on a shoot.
Does it hold up?
Usable locally from day one in ComfyUI with a free browser demo to try first, but heavy: the full split pack is about 66 GB and needs 32 GB of VRAM, and the weights are gated behind an account and licence click-through.
Built with LTX-2.5
- LTX-2.5: Free Open AI Video That Makes the Sound Too - Full Setup Guidedev.to · A full setup guide: the 66 GB split pack, the free API Playground demo, ComfyUI use, and the licence's $10M revenue line.
- LTX 2.5 ComfyUI: VRAM, GGUF, Workflow and Licencevdf.ai · ComfyUI VRAM requirements, community GGUF builds from 8.46 GB, and the licence terms as read in November 2026.
- LTX-2.5 Adds Native Multi-Shot Video and a Diffusion Decoderrits.shanghai.nyu.edu · An explainer of the headline changes: the diffusion video decoder, native multishot, and the robotics fine-tuning checkpoint.
Learn more
First spotted on hf: source.
More AI media tools
Hyperframes: write HTML, render video — HeyGen's open video engine for AI agentsOpen-source renderer from HeyGen that turns HTML, CSS and JavaScript into MP4 video, built so AI coding agents can…media · JEV 0.78MiniMax-H3-Image: open image generation and editing at 16 MP in 6 stepsMiniMax's open-weight image model: text-to-image and editing unified in one 33.1B checkpoint, generating about…media · JEV 0.6FastFrames: Apple's open video models that generate frames in 36 msApple and Cornell's family of autoregressive video diffusion models that generate 480p frames at interactive latency…media · JEV 0.57Genie 4: Google's interactive world model, now in the Gemini APIDeepMind's world model streams continuous 720p interactive worlds at 24fps with native synchronized audio, now…media · JEV 0.54HeyGen Avatar 6: an audio-driven avatar model with script-aware gesturesHeyGen's next-generation audio-driven personal avatar model — script-aware gestures, native 2K output, a personal…media · JEV 0.52Kling Image 4.0: native 8K images with 2,000 characters of clean textKuaishou's Kling Image 4.0 generates natively in 8K, renders up to 2,000 characters of legible text in one image, and…media · JEV 0.48
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.