Grok Imagine 1.5 Lite vs Grok Imagine: Which Should You Use?

TL;DR

• Consider Grok Imagine 1.5 Lite for 1–5-second clips or shapes such as 4:3 and 3:4; consider Grok Imagine for a lower 720p rate or lengths from 6–30 seconds. Lite is $0.043 per second at 720p versus $0.0214 on Grok Imagine. See Lite pricing.
• At a requested 6 seconds and 720p, Lite costs $0.258 and Grok Imagine $0.1284, a $0.1296 gap. A billing difference is not a quality ranking.
• Both are on HiAPI: Lite model page · Grok Imagine model page. Submit through /v1/tasks, read data.taskId, and query the result.

At a glance

This article compares the shared 480p and 720p rates. Set resolution and shape explicitly instead of mixing different defaults.

Two landscape scenes: rice terraces and a baker

Group A: Sunrise balloon over rice terraces

A glass hot-air balloon drifts slowly over a misty valley of terraced rice fields at sunrise. Golden light spills across the water-filled terraces, mist rolls between the hills, a flock of white egrets glides beneath the balloon. Smooth aerial tracking shot, gentle wind sounds and distant birdsong, cinematic, natural colors. No text, no logos.

Both records specify 720p, a requested 6 seconds, and 16:9. Actual output is 1280×720 MP4 at 24fps, with a main video stream of about 6.04 seconds. Both have non-silent audio tracks; that alone does not identify individual wind or bird sounds. This group reuses existing clips and their prompt descriptions. The original Grok Imagine request omitted mode, and the supplied Lite material does not include the complete original task request. Use this group to inspect the scene and composition, not as a comparison with every input parameter matched.

Try Lite → · Try Grok Imagine →

Visible differences: both contain reflective terraces, sunrise light, valley mist, a balloon, and white birds. The Lite balloon and birds occupy more of the frame; the original uses a wider view. These are composition observations, not an overall motion or mist-quality ranking.

Group B: Baker greeting at the oven

Landscape medium shot in a cozy small bakery at morning. A cheerful middle-aged baker with flour on her apron pulls a tray of golden croissants from the oven, looks at the camera and says warmly: "Fresh out of the oven, come try one!" Steam rises, soft window light, natural room tone and oven sounds. No subtitles, no captions, no on-screen text, no logos.

Both use the complete prompt above and request 720p, 6 seconds, and 16:9; Grok Imagine explicitly receives mode=normal. Original outputs are 1280×720 MP4 at 24fps, with container durations of 6.042 and 6.0417 seconds, respectively. Both have non-silent audio; the requested words and lip-sync have not undergone a listening review.

Try Lite → · Try Grok Imagine →

Visible differences: both show a baker holding croissants beside an open oven, with steam rising. The Lite window is behind the baker; the original model’s window is on the left of the frame. Sampled frames spanning each clip show no subtitles or obvious logos. That visual observation does not establish the spoken words or lip synchronization.

Choosing for a workflow

• For budget-sensitive 720p batches, compare Grok Imagine at $0.0214 per second with Lite at $0.043. A thousand requested 6-second tasks differ by $129.60.
• For 1-, 2-, 3-, 4-, or 5-second clips, Lite's duration range covers the request; Grok Imagine starts at 6 seconds.
• For a single clip lasting 16–30 seconds, Grok Imagine's range covers it; Lite tops out at 15.
• For 4:3 or 3:4, use Lite. Both support 16:9, 1:1, 2:3, 3:2, and 9:16.
• For speech or ambience requirements, listen to the output. An audio track does not prove the words are correct or lip-sync is accurate.

Switching: explicit shared fields

{
  "model": "grok-imagine-1.5-lite/text-to-video",
  "input": {
    "prompt": "Landscape medium shot in a cozy small bakery at morning. A cheerful middle-aged baker with flour on her apron pulls a tray of golden croissants from the oven, looks at the camera and says warmly: \"Fresh out of the oven, come try one!\" Steam rises, soft window light, natural room tone and oven sounds. No subtitles, no captions, no on-screen text, no logos.",
    "resolution": "720p",
    "duration": 6,
    "aspect_ratio": "16:9"
  }
}

{
  "model": "grok-imagine/text-to-video",
  "input": {
    "prompt": "Landscape medium shot in a cozy small bakery at morning. A cheerful middle-aged baker with flour on her apron pulls a tray of golden croissants from the oven, looks at the camera and says warmly: \"Fresh out of the oven, come try one!\" Steam rises, soft window light, natural room tone and oven sounds. No subtitles, no captions, no on-screen text, no logos.",
    "resolution": "720p",
    "duration": 6,
    "aspect_ratio": "16:9",
    "mode": "normal"
  }
}

These complete request bodies match the prompt and parameters used for the baker clips shown here. Both explicitly set resolution, duration, and aspect_ratio; send mode only to Grok Imagine. After submission, read data.taskId, query the task, select the type=video result from data.output, and save its content promptly.

Moderation and billing

The Grok Imagine model page warns that results are reviewed after generation and a rejected result can still consume credits. Do not confuse request validation before a task exists with post-generation moderation, or promise that every failure is free.

FAQ

Is Lite always less expensive at 720p?

No. Lite is $0.043 per second versus $0.0214 on Grok Imagine; a requested 6 seconds costs $0.258 and $0.1284, respectively.

Can both produce either a 1-second or a 30-second clip?

No. Lite supports integer seconds from 1–15; Grok Imagine supports 6–30. The shared range is 6–15 seconds.

Does the same 720p label prove matching pixel dimensions?

No. Check shape and actual output dimensions too. Both video pairs here have been checked as 16:9 and 1280×720.

Does a detected audio track prove the requested line was spoken?

No. Volume detection establishes non-silent signal, not words, pronunciation, or lip-sync. Those require listening and visual review.

Is a post-generation moderation rejection always free?

Do not assume so. The Grok Imagine model page warns that rejected generated results can still consume credits; check the rule before submitting.

Next: read the Lite text-to-video API guide, prompt examples, and pricing page, then choose a duration and shape for your task.