• featured
Lightricks/LTX-2.3 Robot

LTX-2.3

LTX-2.3 is a 22B-parameter DiT-based audio-video foundation model designed to generate synchronized video and audio within a single model.

$0.02/s, $0.04/s

Input

Template

Output

[ Go to History Page ]

LTX-2.3 Video Generation

LTX-2.3 is a 22B-parameter video generation model capable of producing high-fidelity videos with synchronized audio from text prompts or reference images. The API is asynchronous — submit a request, receive a job ID, poll for status, and download the result when complete.

Billing

Billed per second of generated video. The tier is determined by resolution.

Tier Resolution Cost
sd ≤ 1280×720 $0.02/s
hd > 1280×720 $0.04/s

The SD/HD boundary is 921,600 total pixels (width × height).

Quick Start

  1. Get an API key (Console → API Keys).
  2. Submit a job: POST /v1/generate with Authorization: Bearer YOUR_API_KEY.
  3. Poll for status: POST /v1/jobs/{id} until status is completed.
  4. Download the video: POST /v1/jobs/{id}/download.

Endpoints

Method Path Description
POST /v1/generate Generate a video (text/image/frame-to-video, auto-detected).
POST /v1/jobs/{id} Check job status.
POST /v1/jobs/{id}/download Download the completed video.

See API: Video Generation and API: Job Management for full endpoint documentation.

Parameters Overview

POST /v1/generate (multipart/form-data)

The type is auto-detected based on which image fields are provided: no images → text-to-video, image/image_uri → image-to-video, first_frame/first_frame_uri → frame-to-video.

Parameter Required Default Description
model Yes — Model identifier. Use "Lightricks/LTX-2.3".
prompt Yes — Text description of the desired video.
image No — Image file for image-to-video.
image_uri No — Image URL or base64 data URI for image-to-video.
first_frame No — First frame file for frame-to-video.
first_frame_uri No — First frame URL or base64 data URI.
last_frame No — Last frame file for frame-to-video.
last_frame_uri No — Last frame URL or base64 data URI.
duration No 5 Length in seconds (4–10).
resolution No — Output resolution as WIDTHxHEIGHT. Takes priority over aspect_ratio.
aspect_ratio No 16:9 Aspect ratio preset (16:9, 4:3, 1:1, 3:4, 4:5, 9:16).
fps No 25 Frames per second. 25 or 50.

Parameter Negotiation

The API never rejects a request due to resolution or duration constraints. Instead, it automatically adjusts parameters to fit within GPU limits, in this order:

  1. FPS — Reduces from 50 to 25.
  2. Duration — Reduces toward 4 seconds.
  3. Resolution — Scales down as a last resort.

When adjustments occur, the response adjustments field lists what changed. If nothing was adjusted, the field is null.

Error Responses

Status Code Description
400 Bad Request Invalid input (e.g., missing prompt, unsupported image format).
401 Unauthorized Invalid or missing API key.
404 Not Found Job ID does not exist or has expired.
409 Conflict Video download requested before generation is complete.
500 Internal Error Internal failure during processing.

All error responses follow this format:

json Copy 
{
  "type": "error",
  "error": {
    "type": "not_found",
    "message": "Job '550e8400-...' not found"
  }
}

Limits

Constraint Value
Duration 4–10 seconds per video
FPS 25 or 50
Aspect ratios 16:9, 4:3, 1:1, 3:4, 4:5, 9:16
Min edge 512 pixels (each dimension)
Max resolution 1920 × 1080
Job expiration 24 hours
Concurrency Queued, one per GPU

Unlock the most affordable AI hosting

Run models at scale with our fully managed GPU infrastructure, delivering enterprise-grade uptime at the industry's best rates.

Contact Sales