Open audio-video generation with LTX-2.3
Lightricks’ DiT foundation model for synchronized video — available in LazyKiwi with promo pricing.
LTX-2.3 is Lightricks’ Diffusion Transformer (DiT) audio-video foundation model (about 22B parameters) designed to generate video with improved prompt adherence and visual quality versus earlier LTX-2 releases. On Hugging Face it ships as open weights under Apache 2.0 with distilled and full variants. In LazyKiwi you can run text-to-video and image-to-video under the LTX workbench profile with resolution and duration caps tuned for production credits. Specs: up to Up to 10s (2160p T2V ≤2s).
- Provider
- Lightricks
- Architecture
- DiT ~22B
- Modes
- T2V / I2V
- License
- Apache 2.0 (HF)
How it works
Use LTX-2.3 in three steps
Open the LazyKiwi workbench, pick LTX-2.3, and generate without leaving the page.
1Choose the model
Select LTX-2.3 from the model list so generation uses the matching upstream endpoint and credit tier.
2Add a prompt or starting frame
Describe motion, camera, and mood — or upload a still for image-to-video when the model supports it.
3Generate and export
Run a small batch, compare takes on mobile framing, then download the keepers for posts, ads, or further edits.
Why LTX-2.3
What LTX-2.3 is built for
LTX-2.3 is available in LazyKiwi with workbench controls aligned to public model capabilities and our production credit table.
Joint video foundation
Built as a modern DiT video stack
On Hugging Face, LTX-2.3 is positioned as a DiT-based audio-video foundation model with distilled (fast) and full/dev checkpoints, plus optional spatial/temporal upscalers for multi-stage pipelines.
- Open weights on Hugging Face
- Distilled + full variants
- Improved prompt adherence vs LTX-2
Creator workflows
Text or image as the starting point
LazyKiwi exposes LTX for both prompt-only clips and image-driven motion, so you can draft a scene from text or animate a still without switching tools.
- Text-to-video drafts
- Image-to-video from a keyframe
- Promo credit tier in workbench
Production limits
Workbench caps you can plan around
LazyKiwi production limits for LTX: T2V up to 480p–2160p (2160p ≤2s, otherwise ≤10s); I2V at 480p/720p ≤10s. No separate audio-drive controls in the current workbench profile.
- T2V multi-resolution path
- I2V 480p / 720p
- Duration capped for credit safety
When to pick it
Fast experimentation at promo pricing
Use LTX when you want open-model motion looks and aggressive promo rates for iteration before upgrading to flagship closed models for final polish.
- 90% OFF credit tier
- Good for A/B draft batches
- Pairs with LazyKiwi templates
Sample directions
Ideas to try with LTX-2.3
Start from these directions in LazyKiwi, then refine prompts to match your brand or story.
Synced product spot
Portrait with natural motion
City night ambience
Specs
LTX-2.3 at a glance
Capabilities exposed through the LazyKiwi workbench (production limits may be narrower than the open-weight ceiling).
- Provider
- Lightricks (HF: Lightricks/LTX-2.3)
- Type
- Video generation (DiT)
- Inputs
- Text prompt; optional image (I2V)
- LazyKiwi modes
- Text-to-video, Image-to-video
- Resolutions
- T2V 480p–2160p; I2V 480p/720p
- Max duration
- Up to 10s (2160p T2V ≤2s)
- Credits
- Promo tier (~10% of peer list price)
- Best for
- Open-model drafts, social hooks, rapid tests
When to use it
LTX-2.3 vs other LazyKiwi video models
LTX is the open Lightricks option with promo pricing. Pick Seedance or Veo when you need longer premium looks.
| Model | Strength | Best for | Pricing |
|---|---|---|---|
| LTX-2.3 | Open DiT video / I2V | Cheap iteration, open-model look | Promo 90% OFF |
| Seedance 2.0 | Smooth social clips | Polished short-form | Standard |
| HunyuanVideo 1.5 | Compact bilingual open video | Efficient T2V/I2V drafts | Promo 90% OFF |
Use cases
What creators build with LTX-2.3
Practical workflows where LTX-2.3 is a strong default inside LazyKiwi.
Social hooks
Draft 5–10s vertical hooks from a prompt or poster still before spending on premium models.
Product motion tests
Animate a packaging or UI still to check motion direction and pacing.
Open-model lookbooks
Explore the LTX aesthetic for campaigns that want open-weight provenance.
Batch ideation
Burn through prompt variants cheaply on the promo credit tier.
FAQ
LTX-2.3 FAQ
What is LTX-2.3?
LTX-2.3 is Lightricks’ open DiT-based audio-video foundation model published on Hugging Face (Lightricks/LTX-2.3), with distilled and full checkpoints under Apache 2.0.
Can I use LTX-2.3 in LazyKiwi?
Yes. Select LTX-2.3 (workbench id ltx) in the video generator for text-to-video or image-to-video within the listed resolution and duration caps.
Does LazyKiwi expose every HF upscaler?
No. Hugging Face lists optional spatial/temporal upscalers for multi-stage pipelines; LazyKiwi currently exposes a production T2V/I2V profile with fixed credit-safe limits.
How is LTX priced?
LTX uses LazyKiwi’s promo video tier: roughly 10% of the cheapest peer list price for the same resolution and duration (shown with a 90% OFF badge).
Try LTX-2.3 in LazyKiwi
Generate with LTX-2.3 in the same workspace as your other image and video models.
More models
Other models available in the same workbench.
