What generation types does the LTX API support?
The LTX API supports the following generation types: text-to-video, image-to-video, audio-to-video, extend, retake, and HDR video-to-video. Each one is a separate endpoint with its own use case and billing model.
What generation types does the LTX API support?
The LTX API is built around video generation. Rather than a single "generate" endpoint, it offers six distinct endpoints — each designed for a specific creative workflow. Choosing the right one depends on what you're starting with and what you're trying to produce.
Text-to-video
Generate a video from a text prompt. This is the most direct endpoint — describe what you want, choose your model, resolution, and duration, and the API returns a video. Billed per second of output video.
Image-to-video
Animate a still image into a video. You provide an image as the first frame, and the API generates motion from it based on your prompt. With LTX-2.3 models, you can also specify a last frame — giving you control over both the start and end of the shot. Billed per second of output video.
→ Image-to-video API reference
Audio-to-video
Generate video driven by an audio track. The API analyses the audio and produces visuals that respond to it — useful for music videos, sound-reactive content, and audio-synced animations. You can optionally provide a reference image for visual direction. Billed per second of input audio. Available on ltx-2-3-pro only.
→ Audio-to-video API reference
Retake
Edit a specific section of an existing video without regenerating the whole thing. You define the start time and duration of the segment to replace, and the API regenerates just that portion — with options to replace video, audio, or both. Billed per second of the input video segment. Available on ltx-2-3-pro only.
Extend
Add more footage to the beginning or end of an existing video. You specify how many seconds to add (between 2 and 20), and the API generates new content that flows naturally from the original. Billed per second of the extended portion plus context frames used from the input. Available on ltx-2-3-pro only.
HDR video-to-video
Convert standard dynamic range (SDR) video to high dynamic range (HDR) output. This endpoint processes an existing video and returns HDR-enhanced frames. It's aimed at post-production and quality-upgrade workflows rather than generation from scratch.
Sync vs async
Every generation type is available in both a synchronous (/v1/) and asynchronous (/v2/) version. The sync version waits for generation to finish and returns the video directly. The async version returns a job ID immediately — you then poll for the result. For production workflows or longer generations, async is recommended.