xAI has moved Grok Imagine Video 1.5, its image-to-video generation model, to general availability. Alongside the developer-facing Imagine API, it is now usable on the web at grok.com/imagine and through the Grok apps for iOS and Android. The biggest changes are speed and audio: the fast tier renders a six-second 720p clip in about 25 seconds and generates sound such as effects and dialogue in the same pass as the video. Elon Musk announced the wider rollout on June 17, 2026.

A Faster Image-to-Video Pipeline With Audio Built In

Grok Imagine takes a single still image and a prompt describing motion, then turns them into a short video clip. Version 1.5 sharply increases generation speed. The fast tier, "Video 1.5 Fast," produces a six-second 720p clip in roughly 25 seconds, down from more than 40 seconds in the previous model — close to half the wait.

Audio is the other headline. Until now, many video generators created a silent clip first and required sound effects or narration to be added in a separate step. Grok Imagine Video 1.5 creates the motion and the audio together within the same generation. Because you can describe the ambience, environmental sound, and character movement inside the prompt, the model reduces the work of bolting audio on afterward.

Faster generation also changes how creators iterate. With less waiting per clip, you can discard weak takes early and try the next one, rather than sitting through a long render only to notice that the motion, timing, or audio is off.

Better Physics, but a 720p Ceiling

xAI says it also tightened the realism of the footage. It reworked how weight and momentum are expressed in motion, reducing the warping of hands, objects, and character movement. Those breakdowns are exactly what tip viewers off that a clip was made by AI.

There is, however, a clear limit. Grok Imagine Video 1.5 outputs up to 720p and does not support 1080p. With rival tools already offering 1080p, that gap is not one that speed and price alone can close. Short social clips and concept drafts are fine at 720p, but for quality-sensitive client work, resolution remains a hurdle to adoption.

Pricing is set relatively low. According to reports, generation costs around 4.20 USD (about 680 yen) per minute of video. Low-cost access becomes a useful comparison point against more capable tools such as Google's Veo and Runway — provided output quality holds up.

※1 USD = 162 JPY (as of June 2026)

New Tools for Production Workflows

Beyond the generation feature itself, xAI is rolling out tooling to organize the work: "Projects" for grouping related outputs, a way to run multiple prompts in parallel, and library search for finding images and videos you have already made.

Rather than waiting for one generation to finish before starting the next, these let creators test several directions at once and quickly retrieve saved material. For short-form work like ads and campaigns, where many variations are compared and narrowed down, that kind of support adds up.

Where It Sits Among the Competition

On the Image-to-Video Arena, an independent ranking that measures human preference through head-to-head comparisons, Grok Imagine Video 1.5 places near the top with an Elo score of about 1,330. Shipping rivals on the same image-to-video field include ByteDance's Seedance 2.0, Kuaishou's Kling 2.6, and MiniMax's Hailuo.

Grok Imagine is a product line that dates back to 2025, and it has previously drawn criticism over the generation of inappropriate imagery based on real people. Version 1.5 is best understood as an update that layers speed, audio, and production tooling onto that line. The biggest remaining task is 1080p support; without it, constraints persist for quality-focused professional use.

Summary

Grok Imagine Video 1.5 brings speed and same-pass audio generation to image-to-video AI, rendering a six-second 720p clip in about 25 seconds. With improved physics and workflow tools that let users test several ideas in parallel, it leans on low pricing to take on rivals. Even so, output tops out at 720p, leaving a quality gap against tools that can produce 1080p. The next question is how far that ease of speed and price can carry it in serious production settings.