xAI's Grok Imagine Video 1.5 Tops Leaderboard and Beats Google Veo

Grok Imagine Video 1.5 is now available on Higgsfield, giving the platform's users direct access to xAI's latest image-to-video model. The model itself launched via the xAI API in preview, and Higgsfield is one of the first creative platforms to integrate it. The timing matters: Elon Musk posted a 40-second AI-generated trailer for The Iliad on X and watched it pull 18.4 million views overnight , the model behind it is Grok Imagine Video 1.5.

What it actually does

grok-imagine-video-1.5-preview turns a single still image into fluid, cinematic video. Give it a starting frame and a prompt describing the motion, and it animates the scene , including camera moves, atmosphere, and physics , while staying faithful to your source image. The model holds detail and lighting from the input frame, so the result continues the original image rather than reinterpreting it.

One important clarification: the official xAI model page explicitly states this preview model currently does not support text-to-video. Every request needs an input image. The broader Grok Imagine API suite supports text-to-video and editing workflows, but grok-imagine-video-1.5-preview is image-to-video only.

The physics layer got a real upgrade

The improvements in 1.5 are specific and testable, not just marketing copy. V1.5 introduces measurable gains in cloth dynamics, water simulation, hair motion, and object interaction. Subject deformation in high-motion scenes is reduced relative to V1.0, with sharper micro-expressions and improved translucent and glass material rendering.

The model runs on xAI's Aurora engine, an autoregressive mixture-of-experts architecture that predicts tokens across interleaved text, image, video, and audio modalities. That unified token stream is what allows audio to generate natively alongside the video in a single pass , no separate audio model, no post-processing step.

Native audio in one shot

Audio generates in the same model pass as video, not as a post-processing layer. Dialogue, sound effects, ambient sound, and background music arrive together with the clip. For production workflows, this removes an entire editing step that every other silent video model requires.

Specs at a glance

• 720p output at 24fps, with clips ranging from 6 to 15 seconds
• Seven aspect ratios including 16:9, 9:16, and 1:1 , both widescreen and full-portrait vertical are supported out of the box
• Accepts JPG, JPEG, PNG, WEBP, GIF, and AVIF input images; outputs H.264 MP4
• Clip extension lets you chain segments into longer multi-shot narratives, with improved continuity between extension boundaries relative to V1.0
• Weaker for long continuous scenes, complex multi-character choreography, and exact frame-by-frame control

Where it sits on the leaderboard

The model launched, entered the Artificial Analysis Video Arena Image-to-Video leaderboard in first place, and sits 52 Elo points above where version 1.0 left off. It arrived at Elo 1404 on the Image-to-Video Arena, placing it above Seedance 2.0, HappyHorse 1.0, and Google Veo in the current standings. Note that Elo scores on human-preference leaderboards shift as more votes come in, so treat the debut rank as a snapshot rather than a fixed position.

How to use it and what it costs

The model is available as a preview through the xAI API and can be set up with just a few lines of code. The official xAI announcement includes a working Python snippet:

import os
import xai_sdk
client = xai_sdk.Client(api_key=os.getenv("XAI_API_KEY"))
response = client.video.generate(
    prompt="Slow cinematic push-in as embers drift across the battlefield",
    model="grok-imagine-video-1.5-preview",
    image_url="https://your-host.com/helmet.jpg",
    duration=10,
    resolution="720p",
)
print(response.url)

A 15-second 720p video costs $2.10 in output charges plus the input. These are competitive rates compared to Midjourney and similar services, but they add up fast in production workflows where you generate at scale. Broader reports put the all-in rate at roughly $4.20 per minute including audio , compare that to Veo and Sora pricing and it looks aggressive. You can also access the model directly on Higgsfield without touching the API.

Where it fits in your workflow

The model is strongest for image-to-video animation of product shots, portraits, and concept frames; social-native short clips where native audio removes post-production overhead; cinematic teaser generation from reference images; and concept testing at speed before committing to a higher-controlled production model.

xAI is now competing directly with video AI providers like Seedance and Google's Veo. Ten months ago xAI had no video product at all. The speed of that trajectory , from zero to leaderboard-topping , is the real story here, and it suggests the gap between frontier video models is narrowing faster than most teams expected.