About LTX 2.5
LTX 2.5 generates video from a text prompt, a first-frame image, or an audio track. A single model covers all three modes.
Two tiers are offered: Fast reaches 4K and 20-second clips; Pro trades resolution (up to 1080p, 10 seconds) for higher fidelity. Both tiers support native multi-shot generation that keeps characters, lighting, and voices consistent across cuts, camera motion effects, and synchronized audio (turn Generate Audio off for silent output).
The open 22B dev checkpoint supports LoRA fine-tuning on video clips with audio.
| Metric | Value |
|---|---|
| Parameter Count | 22 billion |
| Mixture of Experts | No |
Ready to build with LTX 2.5?
Try LTX 2.5 in the Workbench to prompt it, compare outputs, and iterate on prompts without writing any code. When you're ready to ship, call the same model from our API and build your own apps on top of it.
curl -sSf -X POST https://hub.oxen.ai/api/ai/videos/generate \
-H "Content-Type: application/json" \
-H "Authorization: Bearer $OXEN_API_KEY" \
-d '{
"model": "ltx-2-5",
"prompt": "A man making it to the summit on top of a mountain. The man raises his arms in victory and exhales a sigh of relief.",
"video_mode": "fast",
"generate_audio": true,
"aspect_ratio": "16:9",
"resolution": "1080p",
"duration": 6,
"fps": 24,
"camera_motion": "none"
}'Request parameters
Every field LTX 2.5 accepts in the request body. See the API reference for response formats.
https://hub.oxen.ai/api/ai/videos/generate| Field | Type | Required | Default | Description |
|---|---|---|---|---|
prompt | string | Yes | — | Text description of the video to generate, or how to animate the input image. |
video_mode | string | No | "fast" | Fast is quicker and cheaper, with 4K and up to 20 seconds. Pro is higher fidelity, capped at 1080p and 10 seconds.fastpro |
input_image | stringuri | No | — | Optional first-frame image to animate (image-to-video). Leave empty to generate from the prompt alone. |
input_audio | stringuri | No | — | Optional audio track (2-20 seconds) to drive an audio-to-video generation. Output length and lip sync follow the audio; 1080p only. |
tail_image_url | stringuri | No | — | Optional last frame for image-to-video; the model interpolates between the input image and this frame. |
generate_audio | boolean | No | true | Generate synchronized audio for the video. |
aspect_ratio | string | No | "16:9" | 16:99:16 |
resolution | string | No | "1080p" | 720p1080p1440p4k |
duration | integer | No | 6 | Range: 6 – 20 |
fps | integer | No | 24 | Frame rate. 48 and 50 FPS shorten the maximum duration; Pro supports 24, 25, and 50 only.24254850 |
camera_motion | string | No | "none" | Camera motion effect to apply to the generated video.nonedolly_indolly_outdolly_leftdolly_rightjib_upjib_downstaticfocus_shift |
guidance_scale | number | No | — | Audio-to-video only: how strictly the video follows the prompt. Defaults to 5 with a prompt, 9 with an input image. Range: 1 – 15 |