🎬 LTX-2.5 — Text / Image → Video + Audio

A free client demo for Lightricks/LTX-2.5, a 22B-parameter open-weights world model for synchronized video+audio generation. This page runs no model itself (the weights are 70GB+) — every request is sent straight from your browser to the official Lightricks/LTX-2.5 Space and the result is shown here. Since that's a shared public queue, a request can occasionally fail with "no GPU available" — just retry.

Prompt

Starting image (optional — leave empty for text-to-video)

Height 1472

Width 832

Duration (seconds) 2.0

Higher resolution + longer duration need more single-request GPU time — pushing both to max can exceed what your account tier is allowed to reserve on the backend Space.

Decoder

Seed

Auto length Randomize seed Enhance prompt

Getting "quota exceeded"? Use your own HF token for more quota

Hugging Face token (optional)

Anonymous visitors share a small ZeroGPU quota on the backend Space and often hit "no GPU available" / "quota exceeded". Signing in with your own token gives you your own quota instead. Create one (read access is enough) at huggingface.co/settings/tokens. It's kept only in this browser's local storage and sent directly to the official Lightricks/LTX-2.5 Space — never to any other server.

Your generated video will appear here