If you've been watching AI video tools slowly catch up to the hype, LTX Video 2.3 is the version where things start to click. Developed by Lightricks and released in early 2026, it's the first open-source video model that honestly feels production-ready. And with ltx-23.app, you can use it directly in your browser — no GPU, no setup, no headaches.
Here's what's changed, what it does well, and who it's for.
What's New in LTX Video 2.3
This isn't a minor update. Three upgrades stand out.
Sharper visuals. The previous version struggled to preserve fine details — hair, text, edges — once motion kicked in. Version 2.3 rebuilds the video encoder from scratch. The result is noticeably cleaner output with stable motion and far fewer glitchy artifacts.
Native audio. Sound is no longer bolted on after the fact. LTX 2.3 generates synchronized audio in the same pass as the video. Ambient sound, sound effects, and even basic dialogue land on-beat with the visuals. An updated vocoder handles clarity, and lip-sync timing is tight enough for real creative use.
Smarter prompts. The text encoder is four times larger than the previous version. In practice, this means the model actually follows complex instructions on the first try. Camera angle, character movement, scene composition — you can describe it, and it shows up. That alone saves a lot of prompt-rewriting time.
There's also native 9:16 vertical video up to 1080×1920, built specifically for Reels, Shorts, and TikTok. No cropping required.
How the Workflow Feels
The generation flow on ltx-23.app is refreshingly simple. You write a prompt or upload a reference image, dial in your aspect ratio and motion settings, and hit generate. Most clips render in under a minute.
If the clip is too short, the Extend function adds time seamlessly. If one segment doesn't land, Retake lets you fix just that part — without rebuilding from scratch. That iterative control is what separates LTX 2.3 from tools that give you one shot at the result.
Text-to-video and image-to-video both work well. The image-to-video mode is especially solid — it animates a static photo or product shot with believable physics and consistent framing.
Who It's For
LTX Video 2.3 fits a wide range of use cases, but a few stand out.
Marketers can generate ad variants and social creatives at a pace that wasn't possible before. Filmmakers and directors can prototype trailer concepts or pre-visualize scenes before a shoot. E-commerce brands can turn product photos into short demo videos for storefronts and paid campaigns.
For teams that publish consistently, the workflow is fast enough for weekly output without sacrificing quality on paid client work.
Pricing
ltx-23.app runs on a pay-as-you-go credit model. No subscription, no monthly lock-in. Credits never expire.
- Starter — $19.90 for ~18 fast 6-second 1080p clips. Good for testing.
- Creator — $49.90 for ~49 clips. Built for weekly publishing.
- Studio — $99.90 for ~115 clips. Designed for agency-level output.
You start with a free trial before committing to any tier.
The Bottom Line
LTX Video 2.3 closes the gap between "impressive demo" and "actually useful." The visual quality is there. The audio sync is there. The prompt control is there. And with ltx-23.app's cloud interface, you don't need expensive hardware to access any of it.
If you've been waiting for AI video to get serious, this is the version worth trying.