Model release
Articles about Model release.
A 22B Open-Weights Video Model Now Renders Multishot Ads on Hardware You Already Own
LTX shipped a new open-weights release on August 11, 2026, and the genuinely new capability is multishot: one generation that holds character, scene and voice across cuts. Here is what changed, what was already there, and what it costs.
The AI Video Prompt Tricks Seedance 2.5 Retired, and the Four That Still Matter
Ad makers built a folklore of prompt workarounds to stop captions, stray music and drifting actors. ByteDance's 2.5 release note answers several of them inside the model. Here is what is retired, what survives, and what our own probe of the route we call actually enforces.
What is MiniMax H3 (Hailuo 3.0)? 2K video with native audio, explained for ad makers
MiniMax launched H3 on July 31, 2026: one model that generates up to 15 seconds of 2K video with native stereo sound. Here is what those specs actually change for anyone making short-form video ads, and what they do not.
What Is Lyria 3.5? Google's New Music Model and the Ad Soundtrack Gap
Google rolled out Lyria 3.5 in Flow Music on July 29, 2026, with advances in musicality, lyrics, vocals and easier tempo and duration control. Here is what a better music model does, and does not, change for the one layer of an AI ad that still gets outsourced to a stock library.
What Is Flux 3? Black Forest Labs' Multimodal Model, Explained for Ad Makers
Black Forest Labs announced FLUX 3 on July 23, 2026: one model that generates images and up to 20-second video with native audio, and extends to action prediction. It is gated early access, not a launch. Here is what it changes for ad makers, and when you can actually use it.
What Is LTX-2.3? The Open-Weights AI Video Model You Run Yourself
LTX-2.3 is Lightricks' open-weights audio-video foundation model: it generates synchronized sound and picture in one model and runs on your own hardware. Here is what it is, how its architecture works, and how it compares to the models you can run for ads today.
What Is Seedance 2.5? ByteDance's 30-Second AI Video Model
Seedance 2.5 is ByteDance's announced AI video model that generates a 30-second clip in a single pass with up to 50 reference inputs. Here is what it is, what it changes for ad makers, and what to run while it is still pre-launch.