Skip to main content

LTX-2.5 (official API)

First-party LTX API for LTX-2.5 Fast and Pro. Generates video with synchronized audio. OpenTryOn defaults to the async V2 job API (submit → poll → download).

CLI modelAPI modelBest for
ltx-2.5-apiltx-2-5-pro (default) or ltx-2-5-fastHosted T2V / I2V, up to 4K on Fast

Local open weights: LTX-2.5 (local Diffusers) (--model ltx-2.5).

Environment​

export LTX_API_KEY=...          # https://console.ltx.io
# export LTX_API_BASE_URL=https://api.ltx.io

CLI​

opentryon video-generate --model ltx-2.5-api \
  --prompt "A fashion model walking a runway, soft studio light, camera tracking" \
  --model-version ltx-2-5-pro \
  --duration 8 --resolution 1920x1080 --fps 24

# Fast tier + automatic duration
opentryon video-generate --model ltx-2.5-api \
  --prompt "Lookbook: wide establishing shot, then close-up of fabric detail" \
  --model-version ltx-2-5-fast \
  --duration auto --resolution 1280x720

# Image-to-video
opentryon video-generate --model ltx-2.5-api \
  --image look.jpg \
  --prompt "Gentle turn, fabric motion, ambient atelier sound" \
  --duration 8 --resolution 1280x720

Python​

from tryon.api.ltx import LTXVideoAdapter

adapter = LTXVideoAdapter()  # reads LTX_API_KEY
video = adapter.generate_text_to_video(
    prompt="A majestic eagle soaring through clouds at sunset",
    model="ltx-2-5-pro",
    duration=8,
    resolution="1920x1080",
)
open("out.mp4", "wb").write(video)

video = adapter.generate_image_to_video(
    image="look.jpg",
    prompt="Subject turns and smiles, soft breeze",
    duration=8,
    resolution="1280x720",
)

Notes​

  • Resolutions / FPS / duration limits differ by Fast vs Pro — see LTX-2.5 support matrix.
  • --duration auto sends duration: null (automatic length from the prompt).
  • --no-audio disables synced audio.
  • Prefer async (default) for production; set use_async=False on the adapter for the sync V1 endpoints.