MaxVideoAIMaxVideo AI
Log inStart a render

Lightricks model

LTX 2.3 Pro

Generate from text or images, drive timing from uploaded audio, extend existing shots, or retake just one section without leaving the same model page.

Best for teams that want one LTX surface for ideation, sound-led animation, and quick editorial fixes.

Text→VideoImage→VideoAudio→Video (2–20s)Generate: 6–10s4K on generateExtend / Retake from source video
Start generating with LTX 2.3 ProCompare LTX 2.3 Pro vs LTX 2.3 Fast ->

Pay-as-you-go · Price shown before you generate

LTX 2.3 Pro AI video example: A striking woman in a flowing crimson evening gown stands alone on a wet rooftop above a neon city at night...
Audio on10s
  • Price$0.08/s
  • Duration10s
  • Format16:9
View render →

Best use cases

Unified draft-to-edit workflowsAudio-led social performancesImage-to-video with start/end transitionsShot continuation and pickup editsDialogue fixes and selective retakesVertical and landscape campaign variants

Why LTX 2.3 Pro is strategically useful

  • One engine, multiple workflows (Generate, animate from sound, extend existing clips, and retake broken sections without leaving the LTX family.)
  • Audio is no longer just a toggle (You can either generate native audio in standard runs or drive the clip from an uploaded soundtrack in Audio-to-Video.)
  • Better editing leverage (Extend and Retake are practical tools for keeping a good shot instead of starting from zero.)
  • Fits modern content formats (Landscape and vertical generate modes make it easier to plan both hero and social variants.)

Real Specs - LTX 2.3 Pro in MaxVideoAI

The limits that shape your renders.
Price / second1080p $0.08/s1440p $0.16/s4k $0.31/s
Text-to-VideoSupported
Image-to-VideoSupported
Video-to-VideoSupported (extend / retake workflows)
First/Last frameSupported (start + end image in I2V)
Reference image / style referenceSupported (single start image)
Reference videoSupported (source clip for extend / retake)
Max resolution4K
Max duration20s
Aspect ratios16:9 / 9:16
FPS options24 fps / 25 fps / 48 fps / 50 fps
Output formatMP4
Audio outputSupported
Native audio generationSupported
Lip syncSupported
Camera / motion controlsPrompt-based only
WatermarkNo (MaxVideoAI)
Generate workflowsDetails

Text-to-Video and Image-to-Video are the broadest modes: 6 to 10 seconds, 1080p to 4K, vertical or landscape, with native audio available.

  • Use Text-to-Video for net-new story beats.
  • Use Image-to-Video when the first frame must stay on-brand.
  • Add an end image only when the final composition matters.
  • Pick 9:16 when the clip is designed for short-form social.
Audio, extend and retakeDetails

The biggest difference versus simpler engines is not just image quality. It is that LTX 2.3 Pro covers timing-led animation and source-clip edits from the same page.

  • Audio-to-Video turns uploaded audio into the timing spine of the clip.
  • Extend can grow a strong shot at the head or tail.
  • Retake is useful when only one section needs replacement.
  • Context controls help preserve continuity during extension.

LTX 2.3 Pro examples

Local placeholders are used for now on this dev branch. Replace them with final approved LTX 2.3 Pro renders before production launch.

View all LTX 2.3 Pro examples ->

How to Write a Great LTX 2.3 Pro Prompt

Developers guide

LTX 2.3 Pro works best when the prompt matches the active workflow: pure generation, audio-led animation, continuation, or retake.

Tip: duration, resolution, fps and workflow-specific controls live in the UI. Use the prompt for subject, action, camera, lighting, timing intent and what should change.

Quick prompt (fast first pass)

Use a short prompt when you just need to validate the motion idea.

Quick = fast generation drafts.

Template (copy/paste)

[Subject + one action] in [environment], [camera move], [lighting/style], [optional audio cue].

Example

Handheld smartphone UGC clip of a woman unboxing a new skincare bottle at a kitchen table. She peels the seal, smiles, and turns the bottle toward camera. Soft window daylight, natural colors, subtle room tone + packaging crinkle.

Demo prompt - audio-led performance

LTX 2.3 Pro AI video example: A rugged adventurer stands on the edge of a vast desert cliff at golden hour, sunburned face, worn leather...
Audio on10s

A rugged adventurer stands on the edge of a vast desert cliff at golden hour, sunburned face, worn leather jacket, scarf moving in the wind. He opens an ancient brass compass in his hand, and a glowing map made of floating light and dust unfolds above it, illuminating his face and the sand around him. The camera starts wide behind him to show the endless dunes, then pushes in slowly as he raises the compass, ending on a dramatic close view of his eyes reflecting the glowing map. Warm sun, drifting dust, rich bronze textures, epic but clean composition, premium cinematic adventure tone, simple and iconic. Wind over the dunes, faint metallic click from the compass, low atmospheric desert rumble, no text, no logos, no watermark.

View render →

Practical strengths and boundaries

What works best

  • Most complete LTX workflow surface: generate, audio-driven motion, extension and retake in one place.
  • Useful for vertical and landscape planning when the same concept must ship to multiple placements.
  • Extend and Retake reduce waste because you can preserve the strong parts of a clip.

Common problems → fast fixes

  • Audio-to-Video feels visually disconnected -> reduce the scene complexity and let the soundtrack drive the pacing.
  • Extend drifts too far from the original clip -> describe continuity anchors and tighten the new action.
  • Retake changes more than the target section -> shorten the retake window and be explicit about what must stay the same.
  • Image-to-Video feels generic -> make the start frame stronger and simplify the motion request.
  • 4K runs feel expensive for ideation -> validate the motion in a cheaper pass first, then rerun the winner.

Hard limits to keep in mind

  • 4K, 9:16 and fps controls currently apply to the standard generate modes, not every editing workflow.
  • Audio-to-Video, Extend and Retake are standard LTX 2.3 Pro only, not Fast.
  • Retake still needs a precise source window and a clear prompt to avoid drifting too far from the original clip.

Compare LTX 2.3 Pro vs other AI video models

Not sure if LTX 2.3 Pro is the best fit for your shot? These side-by-side comparisons break down the tradeoffs — price per second, resolution, audio, speed, and motion style — so you can pick the right engine fast.

Each page includes real outputs and practical best-use cases.

lightricks

LTX 2.3 Pro vs LTX 2.3 Fast

Generate fast AI video with LTX 2.3 Fast on MaxVideoAI. Text and image workflows support 6–20s clips, 1080p/1440p/4K, native audio, and Fal’s 25/50 fps options.

Compare LTX 2.3 Pro vs LTX 2.3 Fast →

google-veo

LTX 2.3 Pro vs Google Veo 3.1

Generate cinematic 8-second videos with native audio using Veo 3.1 by Google DeepMind on MaxVideoAI. Reference-to-video guidance, multi-image fidelity, pay-as-you-go pricing from $0.52/s.

Compare LTX 2.3 Pro vs Google Veo 3.1 →

openai

LTX 2.3 Pro vs OpenAI Sora 2

Create rich AI-generated videos from text or image prompts using Sora 2. Native voice-over, ambient effects, and motion sync via MaxVideoAI.

Compare LTX 2.3 Pro vs OpenAI Sora 2 →

Safety and likeness guidance

  • Do not generate sexual content or anything involving minors.
  • Do not impersonate real people or public figures without clear authorization.
  • Do not upload private or sensitive personal data in source media.
  • Keep dialogue, likeness and music usage rights clear when using Audio-to-Video or Retake.

FAQ

What is new in LTX 2.3 Pro compared with the older LTX 2.0 pages?

LTX 2.3 Pro adds a unified workflow surface around generate modes plus Audio-to-Video, Extend and Retake, while also supporting 16:9 and 9:16 in the generate flows.

What is the difference between Extend and Retake?

Extend continues a source video before or after the existing footage. Retake replaces a selected section inside the source video, with controls for start time, duration and replacement mode.

Does LTX 2.3 Pro support native audio?

Yes. Text-to-Video and Image-to-Video expose a native audio toggle, and Audio-to-Video uses an uploaded audio file as the driving input.

Does LTX 2.3 Fast include Audio-to-Video, Extend or Retake?

No. LTX 2.3 Fast is limited to text-to-video and image-to-video in the current public Fal routing.