|
TLDR
Grok Imagine Video 1.5 left preview yesterday and is now generally available, with a new Fast version. It tops the image-to-video arena, has native audio, and costs a fraction of the alternatives.
• On June 17, xAI took the model to wide release and retired the preview; it is meant to be built on now.
• A new Fast version makes a six-second 720p clip in about twenty-five seconds, live in the Grok app on web and phone.
• It sits at number one for image-to-video in blind testing, with native audio in the same pass.
• The catches: a 720p cap, quality drop after a few chained extensions, and a rollout still in progress.
|
|
•••
|
|
Grok Imagine Video 1.5 spent the last few weeks as a preview, a developer-only look at xAI's newest video model. Yesterday that changed. xAI took it out of preview into wide release, opened it to everyone, and added a second, faster version alongside it. For anyone making video, the headline is not the model itself, which we looked at when it first appeared. It is that a preview you could not build on is now a tool you can.
The short version: Grok Imagine Video 1.5 turns a still image into a short clip with synchronized audio in a single pass, and as of yesterday it is generally available rather than an experiment. The new Grok Imagine Video 1.5 Fast trades a little quality for speed, making a six-second clip in about twenty-five seconds, and it is live in the Grok app on web and phone. The preview version is now retired.
What makes this worth your attention is the combination: it currently sits at number one for image-to-video in blind user testing, it includes native audio that most rivals still lack, and it costs a fraction of what the alternatives charge. Below is what changed, what it is good at, the catches, and where it fits.
|
|
What Changed
Out of preview, plus a fast mode.
|
|
The core change is status. Until yesterday, the model was a preview through developer access only, the kind of thing you test but do not yet wire into real work. Now it is generally available under a stable name, which means it is meant to be depended on. The preview is discontinued, and anyone building on it moves to the released version.
Alongside it, xAI shipped Grok Imagine Video 1.5 Fast, a high-speed version that makes a six-second clip at 720p in roughly twenty-five seconds, close to twice the speed of the previous model, right inside the Grok app on web, iOS, and Android. xAI also said new workflow features arrive over the next few days: projects to organize your work, several generations running in parallel, and search across everything you have made.
|
|
Why It Matters
Top of the arena, for far less.
|
|
Two things make this more than a routine update. First, quality: in blind user testing on the image-to-video arena, version 1.5 currently sits at number one, ahead of Sora, Veo, Seedance, and Kling, with a clear jump over its own previous version. Second, it generates synchronized audio, sound effects, ambient noise, even lip-sync, in the same pass as the video, which most competing models still cannot do.
Then there is the cost, which is where it gets pointed. Grok Imagine Video 1.5 runs at around four dollars a minute, against roughly thirty for the comparable tier of Sora. A top model with native audio at a fraction of the going rate is the kind of combination that moves people, not because the benchmark is exciting but because the math suddenly works for everyday use.
|
THE NUMBERS
Resolution: 720p at 24 frames per second. Clip length: six to fifteen seconds per pass, extendable by chaining. Audio: synchronized, generated in the same pass. Engine: Aurora, autoregressive rather than diffusion, which keeps faces and camera moves stable across frames. Standing: number one for image-to-video in blind testing, a clear margin over version 1.0. Cost: about four dollars a minute. Fast version: a six-second clip in roughly twenty-five seconds.
|
|
The Catches
Read the limits honestly.
|
|
It is not without constraints, and they are worth knowing before you commit a project to it. The resolution tops out at 720p, while several competitors now reach 1080p, so for large-format or detail-critical work it may not be enough yet. Quality also degrades visibly after two or three chained extensions, and xAI has not given a timeline for a fix, so longer sequences still mean exporting clips and joining them in an editor.
The wider rollout is also still in progress, so exactly what you can reach depends on where you are and which surface you use. None of these sinks it, and xAI's habit of shipping updates every few weeks suggests they will not last, but they are the difference between a tool that fits a job and one that does not.
|
|
What It Means
A real option, not just a benchmark.
|
|
The thing to take from this is not the number-one standing, which will change hands again, as it has several times this year. It is that a genuinely capable video model, with native audio, is now generally available and affordable enough to use for real, not just to admire in a demo. That lowers the bar for making short, sound-on video by a real margin.
So if you make image-to-video, it is worth a fresh look, especially the Fast version for quick iteration. Match it to the job: short, sound-on clips at 720p are its sweet spot, and longer or higher-resolution work still wants chaining or another tool. But the gap between what costs a fortune and what does not just narrowed, and that is the part worth noticing.
|
|
•••
I am putting together a motion pack: image-to-video prompts that actually move, camera directions, motion descriptions, and audio cues tuned for the current video models, each ready to paste onto a still you already have. Built so the clip looks intentional, not just animated.
Want it when it ships? Reply with send me the motion pack and I will get it to you.
|
|
A QUESTION FOR YOU
What still image have you been wanting to set in motion?
Reply and tell me what you would animate first. The still you keep meaning to turn into a clip is the one worth testing a new video model on.
If this was useful, forward it to a creator who has been waiting for video generation to get affordable.
|
|
Until next time,
Luxe Prompting
|
|
Luxe Prompting
AI Image Generation for Creators
|
|
