Grok Imagine Video Learns to Make Its Own Soundtrack

xAI has shipped an update to its image-to-video model, Grok Imagine Video 1.5, and the headline change is one that earlier video generators mostly skipped: sound. A single clip can now arrive with background music, sound effects and lip-synced dialogue baked in, generated together with the picture rather than added in a separate pass afterwards. The audio is meant to stay locked to the motion on screen, so a slammed door lands when the door actually slams.

That matters more than it might sound. For most of the past two years, the impressive AI video demos doing the rounds were silent. Getting believable, synchronized audio out of the same model that draws the frames has been one of the harder problems in the field, and it is the part that separates a striking technical demo from something a creator can actually use.

Faster, and a lot cheaper

The other change is speed. xAI says its Grok Imagine Video 1.5 Fast variant nearly doubles generation speed, rendering a six-second clip at 720p in about 25 seconds, down from roughly 40. The model is available now through grok.com/imagine and the Grok apps on iOS and Android.

Price is where the company is swinging hardest. xAI lists Grok Imagine Video 1.5 at around $4.20 per minute of generated video, and points to OpenAI's Sora 2 at roughly $30 per minute for comparison. If those figures hold up in normal use, that is not a small discount. It is the difference between a tool a hobbyist can play with and one only a studio budget can absorb. Cheap, fast video generation tends to get used in volume, which is its own kind of story.

Read the leaderboard claim carefully

xAI also says the model took the top spot on the Image-to-Video Arena leaderboard, with a jump of 52 Elo points over its predecessor. That claim is worth a grain of salt, not because it is necessarily wrong, but because it comes from the company that built the model, and arena rankings shift as new entrants arrive. The honest summary is that the early head-to-head numbers look strong, and the field moves quickly enough that any number-one badge tends to be temporary.

The bigger picture is the one to watch. Synchronized audio, sub-30-second renders and single-digit dollar pricing are the conditions under which AI video stops being a novelty and starts turning up in real workflows, from social clips to rough animatics. That is the same shift sitting underneath this week's other creative-AI news, including Google's $75 million research deal with the studio A24. The tools are getting good enough, and cheap enough, that the interesting questions are no longer about whether they work. They are about who uses them, and for what.