LTX 2.5
ltx / 2-5-fast-text-to-video

LTX 2.5 is Lightricks' open-source video and audio model family, hosted on Kie.ai across five endpoints in two tiers: LTX 2.5 Fast (text-to-video, image-to-video and audio-to-video) and LTX 2.5 Pro (text-to-video and image-to-video). Text-to-video and image-to-video generate MP4 video with synchronized stereo dialogue, Foley and ambient sound in one pass. Choose Fast for clips up to 20 seconds, 1440p output capped at 10 seconds, or audio-to-video lip sync. Choose Pro for a detail-enhancement pass that delivers sharper faces, textures and short on-screen text in clips up to 10 seconds.

Model Type:
Pricing: LTX 2.5 Fast: 10.8 / 15.6 / 22.8 credits ($0.054 / $0.078 / $0.114) per second of video at 720p / 1080p / 1440p — 40% below official LTX-2.5 Fast pricing. Frame rate does not change the price. High-tier top-ups come with bonus credits that lower the effective price by about 10%.
Input

Describe the scene, action, camera and sound. Put dialogue in quotes; it is spoken in its original language

0/5000

Output resolution (short side): 720p, 1080p or 1440p. Price depends on resolution

Frame rate. 48/50 fps allow up to 10 s

Video length in seconds: 6–20 s in 2 s steps; up to 10 s at 1440p or at 48/50 fps

Aspect ratio of the video

Random seed (0–2147483647). Leave empty for a random seed

Output
output typevideo
Examples

Explore different use cases and parameter configurations

README

Complete guide to using LTX 2.5

LTX 2.5 API for Video and Native Audio in One Pass

Run both LTX 2.5 Fast and LTX 2.5 Pro on Kie.ai. Generate MP4 video with spoken dialogue, sound effects and ambient stereo audio in one pass — choose Fast for clips up to 20 seconds, 1440p output and audio-to-video lip sync, or choose Pro for sharper faces, surface textures and short on-screen text in 720p and 1080p clips up to 10 seconds.
curl https://api.kie.ai/api/v1/jobs/createTask \
-H "Authorization: Bearer $KIE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "ltx/2-5-fast-text-to-video",
"input": {
"prompt": "A baker in a stone bakery taps a warm sourdough loaf beside a crackling oven",
"resolution": "1080p",
"fps": 24,
"duration": 10,
"aspect_ratio": "16:9"
}
}'
Resolution
FPS
Aspect ratio
Duration (s)

Key Capabilities of LTX 2.5

Five KIE hosted endpoints across LTX 2.5 Fast and LTX 2.5 Pro cover text-to-video, first-frame image-to-video and Fast audio-to-video.
Video and Synchronized Stereo Audio in One Pass
01

Video and Synchronized Stereo Audio in One Pass

Across both Fast and Pro text-to-video and image-to-video endpoints, every clip outputs MP4 (H.264 video with AAC stereo audio) with spoken dialogue, Foley effects and environmental sound created alongside the picture — no separate voiceover or sound pass required. Put spoken lines inside quotation marks in your prompt and LTX 2.5 speaks them in their original language, including English, Spanish and Chinese.
LTX 2.5 Pro: Sharper Faces, Textures and Short Text
02

LTX 2.5 Pro: Sharper Faces, Textures and Short Text

Available on ltx/2-5-pro-text-to-video and ltx/2-5-pro-image-to-video, LTX 2.5 Pro adds a detail-enhancement pass on the same model for clearer facial features, skin texture, fabric, metal and glass materials, and short English signs or product labels (keep wording brief and clear; exact spelling is not guaranteed on every run). Pro outputs 720p or 1080p (playground default) for 6, 8 or 10 seconds at 24, 25, 48 or 50 fps with the same stereo audio as Fast.
LTX 2.5 Fast: Up to 20s, or Up to 10s at 1440p / 50 fps
03

LTX 2.5 Fast: Up to 20s, or Up to 10s at 1440p / 50 fps

Built for faster turnaround and longer scenes, LTX 2.5 Fast supports 720p, 1080p (playground default) and 1440p resolutions at 24 (playground default), 25, 48 or 50 fps. In Fast text-to-video and image-to-video, clips run from 6 to 20 seconds in 2-second steps (playground default: 10 seconds) at 720p or 1080p with 24 or 25 fps, or up to 10 seconds at 1440p or at 48/50 fps. Frame rate choice does not change per-second rates.
Multi-Shot Prompts and 20-Second Continuous Takes
04

Multi-Shot Prompts and 20-Second Continuous Takes

Write Shot 1:, Shot 2:, Shot 3: inside a single prompt (tested up to five shots in a 20-second Fast clip) while keeping the same character and outfit consistent from cut to cut. Or describe an unbroken 20-second continuous take in Fast where the soundscape evolves step by step — from fireplace crackle to a whistling kettle and thunder, or from bean grinding to espresso pouring, steam frothing and a saucer clink.
Fast Audio-to-Video: Lip Sync, Singing and Beat-Matched Motion
05

Fast Audio-to-Video: Lip Sync, Singing and Beat-Matched Motion

Audio-to-video is exclusive to LTX 2.5 Fast (ltx/2-5-fast-audio-to-video; not offered in Pro). Provide a public audio URL (MP3, WAV, M4A, AAC, OGG or FLAC up to 50 MB, at least 2 seconds) and a prompt, optionally adding a portrait or 3D character image to generate lip-synced speech or full-verse singing. Without an image, an instrumental track drives beat-matched dance and lighting in 16:9. Your uploaded audio serves as the output soundtrack, and video length follows the audio, capped at 20 seconds (10 seconds at 1440p or 48/50 fps).
First-Frame Image Lock, Five Aspect Ratios and Seeds
06

First-Frame Image Lock, Five Aspect Ratios and Seeds

In both Fast and Pro image-to-video, pass a JPG, PNG or WEBP first frame up to 10 MB to preserve the person, outfit, scene and visual style — across portraits, product photos, anime illustrations and 3D characters — with output aspect ratio locked to your image. Both Fast and Pro text-to-video endpoints support 16:9 (playground default), 9:16, 1:1, 3:2 and 2:3, and all five endpoints accept an integer seed (0 to 2147483647) to reproduce results on identical inputs.
ParameterValues and Endpoint Rules
modelFast: ltx/2-5-fast-text-to-video, ltx/2-5-fast-image-to-video, ltx/2-5-fast-audio-to-video; Pro: ltx/2-5-pro-text-to-video, ltx/2-5-pro-image-to-video
promptRequired on all five endpoints (1 to 5,000 characters); put spoken dialogue in quotation marks; use Shot 1 / Shot 2 for multi-shot cuts
image_urlRequired for Fast and Pro image-to-video, optional for Fast audio-to-video: public JPG, PNG or WEBP URL (max 10 MB) used as the first frame
audio_urlRequired for ltx/2-5-fast-audio-to-video only: public MP3, WAV, M4A, AAC, OGG or FLAC URL (max 50 MB, at least 2 seconds); used as the output soundtrack
resolutionOptional: Fast supports 720p, 1080p (playground default) or 1440p (1440p max 10s); Pro supports 720p or 1080p (playground default)
fpsOptional on all five endpoints: 24 (playground default), 25, 48 or 50 (48 and 50 fps max duration is 10s; fps does not affect per-second rates)
durationOptional: Fast text/image-to-video 6 to 20s in 2s steps (playground default: 10s; max 10s at 1440p or 48/50 fps); Pro text/image-to-video 6, 8 or 10s (playground default: 10s); Fast audio-to-video follows audio_url length, up to 20s (10s at 1440p or 48/50 fps)
aspect_ratioOptional on Fast and Pro text-to-video only: 16:9 (playground default), 9:16, 1:1, 3:2 or 2:3 (image_url locks aspect ratio; Fast audio-to-video without image_url outputs 16:9)
seedOptional integer from 0 to 2147483647 on all five endpoints to reproduce output when rerunning identical inputs
callBackUrlOptional top-level webhook URL for task completion notifications (output MP4 links remain valid for about 14 days)
FeatureLTX 2.5 FastLTX 2.5 Pro
EndpointsText to Video, Image to Video, Audio to VideoText to Video, Image to Video (no Audio to Video)
Resolutions720p, 1080p (playground default), 1440p720p, 1080p (playground default)
Clip duration6–20s in 2s steps (max 10s at 1440p or 48/50 fps); Audio to Video follows audio, up to 20s (max 10s at 1440p or 48/50 fps)6, 8 or 10s
Frame rates24 (playground default), 25, 48, 50 fps24 (playground default), 25, 48, 50 fps
Visual detail & on-screen textStandard detail for multi-shot and long takesDetail pass for sharper faces, skin, fabric/metal/glass textures and short signs or labels
Audio outputMP4 (H.264 + AAC stereo): one-pass dialogue/Foley/ambience, or uploaded audio in Audio to VideoMP4 (H.264 + AAC stereo): one-pass dialogue, Foley and ambient sound (same as Fast)

How to Prompt LTX 2.5

Four practical rules for choosing between Fast and Pro, directing dialogue and sound, and structuring shots.
  1. 1

    Step 1: Choose Between Fast and Pro, Then Pick the Endpoint

    Choose LTX 2.5 Fast when you need clips up to 20 seconds, 1440p output, faster iteration or ltx/2-5-fast-audio-to-video. Choose LTX 2.5 Pro (ltx/2-5-pro-text-to-video or ltx/2-5-pro-image-to-video) for up-to-10-second portrait close-ups, material textures or product shots with brief English signage or labels.
  2. 2

    Step 2: Quote Spoken Dialogue and Describe the Sound

    In text-to-video and image-to-video across both tiers, place exact spoken lines inside quotation marks in the target language, and include a Sound: description of ambient sound, Foley actions and any background instrumental music so audio events align with on-screen motion.
  3. 3

    Step 3: Structure Continuous Takes or Numbered Shots

    For an unbroken take (up to 20 seconds in Fast at 720p or 1080p with 24 or 25 fps, or up to 10 seconds in Pro), describe each action beat and its matching sound in order. For multi-cut storytelling in a single Fast clip, prefix beats with Shot 1:, Shot 2:, Shot 3: and repeat key character clothing details across shots.
  4. 4

    Step 4: Set Resolution, Frame Rate, Seed and Storage

    Pick 720p or 1080p across Fast and Pro (or 1440p in Fast for clips up to 10 seconds) and 24, 25, 48 or 50 fps. Supply an integer seed to reproduce output on identical inputs, and download finished MP4 files to your own storage before temporary URLs expire after about 14 days.

What You Can Build With LTX 2.5

Production workflows across LTX 2.5 Fast and LTX 2.5 Pro.

Portrait, Material and Product Label Close-Ups (Pro)

Use LTX 2.5 Pro for interview and portrait close-ups, beauty, apparel, jewelry or craft material shots, and product packaging or signage scenes with short English text and synchronized sound.

Dialogue-Driven Short Dramas and 20s Takes (Fast)

Use LTX 2.5 Fast to produce 9:16 vertical social clips, multi-shot storyboards and continuous 20-second scenes where voices, Foley and evolving background sound are generated in one pass.

Talking Presenters, Podcasts and Singing Clips (Fast)

Pair a portrait or stylized 3D character image with a voice recording or song verse in ltx/2-5-fast-audio-to-video to create lip-synced hosts, virtual presenters and vocal performances.

Beat-Matched Dance and Style-Locked Animation

Pass an instrumental music track without an image to ltx/2-5-fast-audio-to-video for 16:9 beat-matched dance and lighting, or animate product stills and anime illustrations via Fast or Pro image-to-video.

LTX 2.5 on Kie.ai

01

Quick Start on Kie.ai

Try all five LTX 2.5 Fast and Pro endpoints in the playground on this page, or send a POST request to https://api.kie.ai/api/v1/jobs/createTask with Authorization: Bearer YOUR_API_KEY and model set to ltx/2-5-fast-text-to-video, ltx/2-5-fast-image-to-video, ltx/2-5-fast-audio-to-video, ltx/2-5-pro-text-to-video or ltx/2-5-pro-image-to-video. Poll GET https://api.kie.ai/api/v1/jobs/recordInfo?taskId=... or set callBackUrl to receive your MP4 URL.
03

One Key Across the Kie.ai Market

All models in the Kie.ai market use the same authentication header: Authorization: Bearer YOUR_API_KEY with a key from your Kie.ai API Key Management page.
04

Pay-As-You-Go Credits

Kie.ai is pay-as-you-go with no subscription tier: credits never expire, you can check your balance via the Get Remaining Credits API, and failed tasks are not charged.

Frequently Asked Questions

What is LTX 2.5 on Kie.ai?

LTX 2.5 on Kie.ai provides KIE hosted access to Lightricks' open-source LTX-2.5 video and audio model family across five endpoints: three in LTX 2.5 Fast (ltx/2-5-fast-text-to-video, ltx/2-5-fast-image-to-video and ltx/2-5-fast-audio-to-video) and two in LTX 2.5 Pro (ltx/2-5-pro-text-to-video and ltx/2-5-pro-image-to-video). Text-to-video and image-to-video generate MP4 (H.264) video and AAC stereo audio together, while Fast audio-to-video synchronizes video to your uploaded audio.

What is the difference between LTX 2.5 Fast and LTX 2.5 Pro?

LTX 2.5 Pro adds a detail-enhancement pass on the same model for sharper faces, skin, fabric/metal/glass textures and clearer short on-screen text in 720p or 1080p clips of 6, 8 or 10 seconds (text-to-video and image-to-video only). LTX 2.5 Fast generates faster, runs up to 20 seconds at 720p or 1080p (24/25 fps), adds a 1440p option (up to 10 seconds), and includes audio-to-video. Stereo audio quality is identical across both tiers.

Does LTX 2.5 generate spoken dialogue, sound effects and on-screen text?

In both Fast and Pro text-to-video and image-to-video, spoken dialogue, Foley sound effects and ambient audio are generated with the picture in one pass; put spoken lines in quotation marks to have them spoken in their original language. For scenes with short English signs or product labels, use LTX 2.5 Pro (keep wording brief and clear, as exact text rendering is not guaranteed every run; Fast is not intended for on-screen text).

What resolutions, frame rates and durations do Fast and Pro support?

Both tiers support 24 (playground default), 25, 48 and 50 fps. LTX 2.5 Fast supports 720p, 1080p (playground default) and 1440p, with text/image-to-video durations from 6 to 20 seconds in 2-second steps (capped at 10 seconds at 1440p or at 48/50 fps). LTX 2.5 Pro supports 720p and 1080p (playground default) with durations of 6, 8 or 10 seconds.

How do multi-shot prompts work in LTX 2.5?

Label sequential camera cuts inside a single prompt as Shot 1:, Shot 2:, Shot 3: (tested up to five shots in a 20-second Fast clip) and repeat key subject and outfit details so the same character stays consistent across shots.

How does audio-to-video work, and is it available in LTX 2.5 Pro?

Audio-to-video is available only in LTX 2.5 Fast (ltx/2-5-fast-audio-to-video). Pass a public audio_url (MP3, WAV, M4A, AAC, OGG or FLAC, at least 2 seconds and up to 50 MB) plus a prompt; image_url (JPG, PNG or WEBP up to 10 MB) is optional. Your uploaded audio becomes the output soundtrack: with image_url the video matches the image's aspect ratio for lip-synced speech or singing, and without image_url it outputs 16:9. Video length follows the audio, up to 20 seconds (capped at 10 seconds at 1440p or 48/50 fps).

How long do generated LTX 2.5 video links remain valid?

Result MP4 URLs remain valid for approximately 14 days. Download and transfer completed videos to your own storage for long-term retention.

How do I call the LTX 2.5 API on Kie.ai?

Create an API key on the Kie.ai API Key Management page, then send a POST request to https://api.kie.ai/api/v1/jobs/createTask with model set to one of the five LTX 2.5 Fast or Pro model IDs and your parameters under input. Poll GET https://api.kie.ai/api/v1/jobs/recordInfo?taskId=... or pass callBackUrl to receive the MP4 link. See the Fast Text to Video, Fast Image to Video, Fast Audio to Video, Pro Text to Video and Pro Image to Video API docs.

How are LTX 2.5 Fast and LTX 2.5 Pro billed on Kie.ai?

Both LTX 2.5 Fast and LTX 2.5 Pro are billed per second of video according to tier and resolution (720p, 1080p, or Fast 1440p), with identical per-second rates across the endpoints within each tier and no surcharge for 25, 48 or 50 fps. Text-to-video and image-to-video bill by your selected duration, Fast audio-to-video bills by output video length, and failed tasks are not charged. Current per-second rates for Fast and Pro are shown on the pricing card on this page.