LTX-2.5 Fast
Text to video · Image to video
Starting price $0.1427 / second
By Lightricks
LTX-2.3 22B is Lightricks' video model with every sampling control exposed: video and sound from a prompt or an image, with separate guidance for picture and audio.
ltx-2.3-22b/text-to-videoThis model has other modes with their own ids: ltx-2.3-22b/image-to-video.
Quote the input, start the generation with the quote, then poll the job. Set RAYWAKE_API_KEY first; see Authentication.
# 1. Price the exact input (quotes are valid for up to 10 minutes)
curl -s https://api.raywake.com/v1/quotes \
-H "Authorization: Bearer $RAYWAKE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "ltx-2.3-22b/text-to-video",
"input": {
"prompt": "A blacksmith hammers a glowing blade on an anvil in a dark forge. Sparks scatter with every strike, the hammer rings out and the fire roars behind him."
}
}'
# 2. Start the generation with the quote_id from step 1
curl -s https://api.raywake.com/v1/generate \
-H "Authorization: Bearer $RAYWAKE_API_KEY" \
-H "Content-Type: application/json" \
-H "Idempotency-Key: $(uuidgen)" \
-d '{"model": "ltx-2.3-22b/text-to-video", "quote_id": "QUOTE_ID", "input": {"prompt":"A blacksmith hammers a glowing blade on an anvil in a dark forge. Sparks scatter with every strike, the hammer rings out and the fire roars behind him."}}'
# 3. Poll until the status is terminal
curl -s https://api.raywake.com/v1/jobs/JOB_ID \
-H "Authorization: Bearer $RAYWAKE_API_KEY"| Name | Type | Description |
|---|---|---|
promptrequired | string | The prompt to generate the video from. |
num_frames | integer | The number of frames to generate. |
video_size | string (enum) | The size of the generated video. |
generate_audio | boolean | Whether to generate audio for the video. |
use_multiscale | boolean | Whether to use multi-scale generation. If True, the model will generate the video at a smaller scale first, then use the smaller video to guide the generation of a video at or above your requested size. This results in better coherence and details. |
fps | number (enum) | The frames per second of the generated video. |
num_inference_steps | integer | The number of inference steps to use. |
video_cfg_scale | number | The Classifier-Free Guidance (CFG) scale for the video. Higher values result in more consistent and focused video content. |
video_stg_scale | number | The Spatiotemporal Guidance (STG) scale for the video. Higher values result in more consistent and focused video content. |
video_rescaling_scale | number | The rescaling scale for the video. Controls the ratio between classifier-free guidance and spatiotemporal guidance. |
video_modality_scale | number | The modality scale for the video. Controls the ratio between video and audio modalities. |
audio_cfg_scale | number | The Classifier-Free Guidance (CFG) scale for the audio. Higher values result in more consistent and focused audio content. |
audio_stg_scale | number | The Spatiotemporal Guidance (STG) scale for the audio. Higher values result in more consistent and focused audio content. |
audio_rescaling_scale | number | The rescaling scale for the audio. Controls the ratio between classifier-free guidance and spatiotemporal guidance. |
audio_modality_scale | number | The modality scale for the audio. Controls the ratio between video and audio modalities. |
gradient_estimation_gamma | number | The gamma of gradient estimation during denoising. Set to 0 to disable. |
use_restart_sampling | boolean | Whether to use restart sampling. This will inject a small amount of noise during each denoising step, which can help improve the quality of the generated video. |
scheduler | string (enum) | The scheduler to use. |
acceleration | string (enum) | The acceleration level to use. |
camera_lora | string (enum) | The camera LoRA to use. This allows you to control the camera movement of the generated video more accurately than just prompting the model to move the camera. |
camera_lora_scale | number (enum) | The scale of the camera LoRA to use. This allows you to control the camera movement of the generated video more accurately than just prompting the model to move the camera. |
negative_prompt | string | The negative prompt to generate the video from. |
seed | integer | The seed for the random number generator. |
enable_prompt_expansion | boolean | Whether to enable prompt expansion. |
video_output_type | string (enum) | The output type of the generated video. |
video_quality | string (enum) | The quality of the generated video. |
video_write_mode | string (enum) | The write mode of the generated video. |
sync_mode | boolean | If `True`, the media will be returned as a data URI and the output data won't be available in the request history. |
distill_lora_first_pass_scale | number | The scale of the distill LoRA to use for the first pass. Set to 0 to disable. |
distill_lora_second_pass_scale | number | The scale of the distill LoRA to use for the second and subsequent passes. |
A succeeded job lists its results under outputs, stored on our side. See Jobs for every field.
"outputs": [
{ "type": "video", "url": "https://…", "mime_type": "video/mp4", "thumbnail_url": "https://…" }
]The exact charge for your input comes from a quote. How charges settle
Fields marked * are required. Review the price before generating.
LTX-2.3 22B sets length by frame count, from 9 to 481 frames, and size by preset: square, square HD, or 4:3 and 16:9 in landscape or portrait. Sound is on by default and comes from the same prompt.
Every sampling setting is open: steps, scheduler, guidance for video and audio, multi-scale rendering and prompt expansion. You can save as MP4, WebM, ProRes 4444 MOV or GIF. The image mode adds a strength for how closely the clip keeps your picture.
Check sound timing, repeated textures and object continuity across the entire clip.