xAI's Grok Imagine 1.5 Preview Takes Top Spots in Video AI Benchmarks
The model's benchmark performance shows xAI has closed the gap with leading video generation tools, particularly in preserving input frame details and following motion prompts.
Reporting from 1 sources: GIGAZINE.
xAI released Grok Imagine 1.5 Preview on June 3, 2026, a video generation AI that turns a single image into a 720p, 15-second clip. It ranked second in the Video Arena with audio and first in the Design Arena's Image to Video category, beating Seedance 2.0 and Google's Veo 3.1 in benchmark comparisons.
Grok Imagine 1.5 Preview generates video from a still image and a text prompt, keeping the original frame's lighting and detail rather than reinterpreting it. The model outputs up to 720p resolution and 15 seconds of footage. It is available as a preview through the xAI API, with a cost of $8.40 per minute of generated video.
In the Video Arena benchmark by Artificial Analysis, which evaluates image-to-video generation quality, Grok Imagine 1.5 Preview placed second with audio and third without audio. In the crowdsourced Design Arena, it scored an Elo of 1357 in the Image to Video category, surpassing Seedance 2.0 for first place. Example comparisons show Grok Imagine 1.5 Preview handling zipper motion and mirror reflections more accurately than Google's Veo 3.1.
Synthesized by Yomimono from the 1 cited source below, including Japanese-language reporting where cited, then editorially reviewed before publishing.