xAI launched Grok Imagine Video 1.5 Preview ahead of this June 3, 2026 report. Its image-to-video ranking was a launch-era result. The current official model page names grok-imagine-video-1.5, retains the preview alias, and lists updated resolution, pricing and rate-limit details below.
Access: this article covers developer APIs and provider-specific image-to-video offerings. An API listing does not establish a free consumer rollout or eligibility for a particular X subscription. Check the service and account you intend to use; this update did not sign in or generate a video.
An image-to-video preview workflow with synchronized audio is documented in the Replicate readme; Kie.ai also describes native audio generation. Those provider descriptions do not establish a measured quality advantage over Sora, Runway or Kling, and their workflow limits should not be assumed to match every direct xAI endpoint.
Current official model name: grok-imagine-video-1.5
Aliases: grok-imagine-video-1.5-preview and grok-imagine-video-1.5-2026-05-30
Modalities listed by xAI: Text, Image and Audio inputs; Video output.
Keep provider workflows separate: Replicate’s current preview README requires an input image for every prediction and directs text-to-video users to xai/grok-imagine-video. The current xAI model page lists text, image and audio inputs; it no longer contains the quoted text-to-video prohibition. That modality list alone does not qualify an exact request schema. Use the documentation for your chosen endpoint rather than transferring Replicate’s preview restrictions to all xAI access.
The benchmark: #1 on the Image-to-Video Arena
The following table preserves this article’s June 3, 2026 launch-era ranking snapshot for the arena.ai/leaderboard. The current full board and each uncertainty interval have not been reverified in this update; these are not today’s standings.
Rank
Model
Elo
1
grok-imagine-video-1.5-preview-720p
1473 ±9
2
dreamina-seedance-2.0-720p
1467 ±11
3
happyhorse-1.0
1443 ±12
4
grok-imagine-video-720p (the 1.0 predecessor)
1421 ±6
5
veo-3.1-audio
1397 ±11
Kie.ai repeats the launch-era scores 1473 for Grok and 1467 for Seedance, with a reported 52-point improvement; Oimi AI reported the result on June 1, 2026. The article’s original arithmetic, 1473 − 1421 = 52, is consistent, but it is not a current ranking measurement.
Different boards and dates: the original report mentioned conflicting launch scores and an Artificial Analysis snapshot led by Seedance 2.0. Neither that snapshot nor the current Artificial Analysis I2V leaderboard was fully qualified for this update. The historical reports explain the launch attention; they do not prove that Grok remains #1 today.
Verified specs
The table separates direct xAI model-page information from provider-specific preview documentation. Duration, aspect ratio and workflow settings belong to the chosen provider; this update did not exercise the API.
Spec
Value
Source
Workflow
Replicate preview: image-to-video only, input image required; do not generalize to every xAI endpoint
Kie’s form lists JPEG, PNG, WEBP and JPG; fal’s full current input-format schema was not qualified here
fal
Audio
Native, generated & synchronized in one pass
Replicate / Kie.ai
Status
Replicate labels its offering preview; xAI’s current model name omits the preview suffix and keeps it as an alias
xAI Docs
Pricing (this is the part to plan around)
The current official xAI model page lists per-second video output charges by resolution and $0.01 per input image. Prices below refer to direct xAI, not a reseller’s separate bill.
Item
Price
Output — 480p
$0.08 / second
Output — 720p
$0.14 / second
Output — 1080p
$0.25 / second
Input image
$0.01 each
Rate limit
10 requests / second on the current model page
Region (official)
us-east-1
Direct xAI examples below include one $0.01 input image, using the listed output rate. A ten-second 1080p clip is $2.50 output + $0.01 image = $2.51.
These calculations use the stated output and image rates. The official page separately says preset-voice audio input is free; that should not be generalized into every provider’s audio-billing policy. Check fal pricing or your chosen reseller’s price before sending a request.
How to use it (API only)
Choose a provider and confirm its current request schema, account permissions and billing before generating. The following is documentation-based guidance; no request was executed for this update.
1. Direct via xAI (api.x.ai): the official page lists grok-imagine-video-1.5 with the preview and dated aliases. It lists 480p, 720p and 1080p pricing. Follow the xAI Video Generation / Image-to-Video docs and its linked generation/REST references for the exact endpoint and required fields; a provider’s 1–15-second image-to-video schema is not automatically the direct xAI schema.
2. Via a reseller: the following are provider listings. Playground access, free credits and billing depend on the provider and account; they were not tested here.
For example, Kie’s form currently exposes prompt, image URLs, duration, aspect ratio, resolution and a configurable checker. Replicate’s README supplies its own preview limits. Read the chosen schema; this article does not provide an executed or verified JSON request.
Prompting tips (from xAI’s own Replicate guide)
The model already sees your image, so prompt for motion, not description (Replicate prompt guide):
Don’t re-describe the image — tell it what should change (action, camera move, atmosphere).
Don’t contradict the source — match the prompt to what’s actually in the photo.
Be specific about motion intensity — “car racing past at high speed” beats “car passing.”
Always give camera direction — pan, tilt, dolly, orbit, slow push-in, handheld, etc.
Negative prompts are ignored — describe what you want instead.
Add an AUDIO: block at the end to steer sound design (music, SFX, ambient, short dialogue).
Shorter is more stable — 5–8s is the sweet spot; 15s works but is more prone to artifacts.
Honest limitations
Resolution depends on the provider: xAI now lists 1080p at $0.25 per output second; Replicate’s preview README still lists a 720p maximum.
Quality degrades after ~2–3 chained extensions when stitching clips into longer narratives (community-reported; no xAI fix timeline). (JXP)
Replicate continues to label its offering preview. The direct xAI model name and aliases differ; the page alone does not establish identical release status or behavior across providers.
Weaker for long continuous scenes, complex multi-character choreography, and exact frame-by-frame control. (ImagineGo review)
Claims I could NOT fully verify (treat with caution)
The following were unverified claims in the original launch report. They are retained as historical cautions, not current facts or Kingy test results.
Generation speed: “~17 seconds” (JXP) vs “~20–30 seconds for a 5s 720p clip” (Basenor). Not independently confirmed.
“2–3× faster than Seedance 2.0” — Basenor claim only.
Infrastructure: “Aurora autoregressive MoE engine, trained on Colossus 2 with ~555,000 GPUs” (Basenor) vs “Aurora trained on 110,000 NVIDIA GB200 GPUs” (JXP). These conflict and neither is in the official model docs.
“Hotshot acquired March 2025” and “1.245 billion videos generated in January 2026” — single-source, unverified.
The @grok X announcement — could not load (login wall), so its specific wording is unconfirmed.
Signup form provided by Beehiiv. After submitting, check your inbox for "Confirm your subscription to The Kingy Brief" and open its confirmation link. Check Spam or Promotions if you cannot find it.
Fridays at 09:00 Vancouver time: source-checked AI changes, original tests and one practical thing to try. We skip a week when there is not enough verified material. Free. Unsubscribe anytime.
