You take a photo you already have, type a sentence about how it should move, and a few seconds later you get back a short video with sound that fits the action. Footsteps land when feet hit the ground, wind hums under an outdoor shot, a character's lips move in time with the words. That is the pitch behind Grok Imagine Video 1.5, the new model xAI started rolling out to regular people this week on grok.com and in the Grok app.
What actually shipped
Grok Imagine 1.5 turns a still image into a short video. You give it a starting picture, describe the motion, the camera move, and the mood you want, then pick the resolution and length. It animates the picture from there. There is also a faster version called Video 1.5 Fast, which is the one most people will use on the web and in the phone apps, with the full model available to developers through xAI's programming tools.
The headline change is the sound. xAI says sound effects, background ambience, and dialogue are now generated in the same pass as the picture, and they land on the action instead of arriving as a separate track you have to add yourself. Speech is clearer and the lip movement matches it more closely. In the old way of working, your video came out silent and you went hunting for the right audio afterward, then nudged it into place clip by clip. Here the first thing the model hands you already has sound on it.
It is also quicker. xAI says the Fast version makes a six-second, 720p clip in about 25 seconds, down from 40 seconds or more on the previous model. The motion holds together better too, with fewer of the warps and rubbery movements that gave earlier AI video away.
Why people are paying attention
Grok Imagine 1.5 tops the Image-to-Video Arena, a public leaderboard where people vote on clips without knowing which model made them. That blind voting matters, because nobody can game it by recognizing a favorite brand. In that test it edged past Seedance 2.0, ByteDance's model that a lot of creators had been calling the best image-to-video tool available. Coming out ahead of the one everyone was already pointing to is a real result.
The timing helps. OpenAI shut down its standalone Sora app earlier this year, which left a lot of creators shopping for the next thing they could open and start making clips in. Grok showed up fast, cheap, and already inside an app many of them have. The demos have backed up the hype: one creator, who goes by @heavypulp, built a full Trojan War trailer styled after the Iliad entirely in Grok Imagine 1.5, and xAI features it as proof of what the model can do in real hands.
On price, this is the cheap end of the market. For developers using it through the programming tools, Grok runs at roughly $4.20 a minute of 720p video, which works out far below what comparable services charge. For everyone else, it is available to use on grok.com and in the Grok app, with a paid SuperGrok plan at $30 a month for people who want higher limits.
The honest catch
The reason Grok is the cheap, fast option is that it does not aim for the high end. It caps out at 720p, which is fine on a phone screen and on most social feeds but short of the 1080p that Google's Veo and other premium models produce. The clips are short, up to about 15 seconds. There are fewer fine-grained controls over the motion, and since the model is new, there is not yet a deep well of community know-how on how to prompt it well.
So the way to think about Grok Imagine 1.5 is as a draft tool. It gets you to something watchable in seconds, for almost nothing, with the sound already on it. When you need a polished, broadcast-grade final cut, a model like Veo is still the one you reach for to finish.
If you run a small business or make things for social media, this changes what is realistic to attempt without a camera crew or an editor. A single product photo can become a moving ad. A logo can get a slow push-in with a bit of ambient sound under it. A still you shot on your phone can carry a few seconds of motion and a voice. Because the sound is generated in the same pass, the clip you get back is closer to something you could post than to step one of a four-step editing job, which is where most people used to give up.
The honest limit keeps it useful instead of overhyped. At 720p and a few seconds per clip, this is built for a quick social post, a product teaser, or a rough version of an ad concept you want to show a client before committing real money to it. It is not your finished commercial. Treat it as the cheapest, fastest way to find out whether an idea is worth shooting for real, and you will get a lot out of an afternoon of playing with it.