Text-to-Video
Create dynamic videos from text prompts. Generate engaging visual content without cameras or editing software.
- Text-to-Video
- Image-to-Video
- Audio-to-Video
LTX-2.5 22B Distilled INT8
- Free tier available
- No subscriptions
- No credit card required
Capabilities
What you can do with LTX-2.5 22B Distilled INT8.
Technical Specifications
Resolution
512–1344 x 512–1344 px
Default Size
768 x 768 px
Inference Steps
8 steps (default: 8)
Guidance
Not required
Video Length
49–241 frames at 24 fps (2–10s)
How to use in deAPI
Integrate LTX-2.5 22B Distilled INT8 into your app with a single API call.
POST https://api.deapi.ai/api/v2/videos/generations
curl · deAPI videos/generations
curl -X 'POST' \ 'https://api.deapi.ai/api/v2/videos/generations' \ -H 'accept: application/json' \ -H 'Authorization: Bearer YOUR_API_KEY' \ -H 'Content-Type: application/json' \ -d '{ "model": "Ltx2_5_22B_Dist_INT8", "prompt": "a cat walking through a garden", "width": 768, "height": 768, "frames": 65 }'{ "model": "Ltx2_5_22B_Dist_INT8", "prompt": "a cat walking through a garden", "width": 768, "height": 768, "frames": 65 }
Tip: The API returns a request_id. Use webhooks (recommended) or poll GET /jobs/{request_id} for results.
POST https://api.deapi.ai/api/v2/videos/animations
curl · deAPI videos/animations
curl -X 'POST' \ 'https://api.deapi.ai/api/v2/videos/animations' \ -H 'accept: application/json' \ -H 'Authorization: Bearer YOUR_API_KEY' \ -F 'model=Ltx2_5_22B_Dist_INT8' \ -F 'prompt=gentle camera zoom, cinematic lighting' \ -F 'first_frame_image=@/path/to/file' \ -F 'frames=65'{ "model": "Ltx2_5_22B_Dist_INT8", "prompt": "gentle camera zoom, cinematic lighting", "first_frame_image": "@/path/to/image.png", "frames": 65 }
Tip: The API returns a request_id. Use webhooks (recommended) or poll GET /jobs/{request_id} for results.
POST https://api.deapi.ai/api/v2/videos/audio-syncs
curl · deAPI videos/audio-syncs
curl -X 'POST' \ 'https://api.deapi.ai/api/v2/videos/audio-syncs' \ -H 'accept: application/json' \ -H 'Authorization: Bearer YOUR_API_KEY' \ -F 'model=Ltx2_5_22B_Dist_INT8' \ -F 'prompt=a person speaking to the camera' \ -F 'audio=@/path/to/file'{ "model": "Ltx2_5_22B_Dist_INT8", "prompt": "a person speaking to the camera", "audio": "@/path/to/audio.mp3" }
Tip: The API returns a request_id. Use webhooks (recommended) or poll GET /jobs/{request_id} for results.
Need an API key?
Sign up for free and get $5 in credits to start.
Similar Models
Other models with similar capabilities you might find useful.
LTX-Video 13B
Up to 768×768 resolution. 30–120 frames at 30fps (1–4 sec). Text-to-video and image-to-video modes. Single-step distilled inference.
Image-to-Video
Text-to-Video
LTX-2.3 22B
State-of-the-art video generation model with 22 billion parameters. Supports text-to-video, image-to-video, and audio-to-video with high-quality cinematic output.
Image-to-Video
Text-to-Video
Audio-to-Video
MiniMax H3 33B Turbo INT8
Omni-modal 33B video model that generates video and synchronized stereo audio in a single pass. Supports text-to-video and first/last-frame image-to-video, with spoken dialogue in...
Image-to-Video
Text-to-Video
Frequently Asked Questions
LTX-2.5 22B is Lightricks' open-weights audio-video model with 22 billion parameters. It generates video and synchronized sound in a single pass and supports three modes on deAPI: text-to-video, image-to-video, and audio-to-video lip-sync.
Four things matter most. Native synchronized audio, so a clip arrives with its own ambience, effects, music, or dialogue. Native multishot generation, so one prompt can contain several shots joined by explicit cuts. A new diffusion video decoder that replaces the VAE stage, giving sharper faces, textures, and on-screen text. And a custom Gemma 4 12B text encoder that holds long, detailed prompts together instead of losing detail toward the end.
Yes. Audio is generated jointly with the picture rather than added afterwards, so footsteps, speech, and impacts land on the frames that show them. Describe the sound you want in the prompt — ambience, music, and dialogue — and put spoken lines in quotation marks.
Multishot lets one generation contain several distinct shots joined by cuts, instead of a single continuous take. Write the scene as one chronological paragraph and name each transition in prose, for example "a hard cut transitions to a close-up of...". Re-establish the framing after every cut, reuse the same visual identifiers for recurring people, and say whether the music or dialogue continues across the cut. Two to four shots per generation works best.
Provide a portrait image and an audio clip. LTX-2.5 generates a video in which the person in the image speaks the audio with synchronized lip movements, following your prompt for framing, expression, and delivery. It is the fastest route to AI avatars, presenter clips, and localised talking-head content.
Write one flowing paragraph in the present tense, roughly four to eight sentences for a single shot. Cover six things: the shot scale, the scene's lighting and palette, the action as a natural sequence, the characters described through physical cues rather than abstract emotions, the camera movement, and the audio. Start simple and layer detail as you iterate. Lightricks publishes the full guide at docs.ltx.video.
Sign up for a free deAPI account at https://deapi.ai/register and get $5 in free credits, no credit card required. Then open the playground, select LTX-2.5 22B, and generate your first result in seconds.
- No subscription
- No credit card required
- Pay only for what you use
LTX-2.5 22B Distilled INT8 for Free
Get $5 in free credits and start generating with LTX-2.5 22B Distilled INT8.