# Grok Imagine

Grok Imagine in Sloosh: xAI reference-to-video with sound, 4 to 10 seconds at up to 720p, in seven aspect ratios, from up to 7 images.

Last updated: 2026-10-05

**Key facts**

- Grok Imagine is xAI's reference-to-video model: it keeps a person, product or outfit from up to 7 reference images without fixing the first frame.
- Grok Imagine makes 4, 6, 8 or 10 second clips in Sloosh at 480p or 720p.
- Grok Imagine offers seven aspect ratios in Sloosh: 1:1, 16:9, 9:16, 4:3, 3:4, 3:2 and 2:3.
- Grok Imagine always makes sound, and accepts prompts of up to 5,000 characters in Sloosh.

Web page: https://sloosh.ai/docs/grok-imagine

Grok Imagine makes short clips that keep a person, product or outfit from your reference images, without fixing the first frame. It always makes sound and offers seven aspect ratios.

## When to use it

- A quick, cheap clip that keeps a product consistent.
- Square, portrait or 3:2 clips that many video models don't offer.
- One clear action per clip; several actions can glitch.

## How to prompt it

- Open with the subject and action, in the order they happen.
- Name the camera move: push-in, orbit, pan, tracking.
- Say what to keep from the reference: "Keep the sneaker's colours exactly as in the reference."
- End with a "Sound:" line and describe sounds by material and place.
- xAI publishes no official prompting guide; this advice comes from xAI's reference docs and community testing.

## Example

One product reference, one action, then the sound.

Input: product photo: ![A white mesh running sneaker with sage-green panels and a thick white sole, side view on light grey](https://sloosh.ai/docs/in-sneaker.webp)

->

Result: [![A runner's hand picking up the sneaker from a wooden park bench](https://sloosh.ai/docs/m-grok-imagine.webp)](https://sloosh.ai/docs/m-grok-imagine.mp4) (video)

_16:9, 720p, 6 seconds, Enhance off. Clips here play without sound; press one to pause._

**Prompt**

```text
The sneaker from Image 1 sits on a sunlit wooden bench in a park, then a runner's hand picks it up and laces it. Keep the sneaker's white mesh, sage-green panels and white sole exactly as in the reference. Slow push-in at low height, warm morning light, shallow depth of field.
Sound: birdsong, laces pulled tight, a light breeze in the leaves.
```

## Settings in Sloosh

- Aspect ratio: 1:1, 16:9, 9:16, 4:3, 3:4, 3:2 or 2:3.
- Resolution: 480p or 720p.
- Length: 4, 6, 8 or 10 seconds.
- References: up to 7 images.
- Start frame: optional; no end frame.
- A start or end frame can't be combined with reference images.
- Sound: always on.
- Prompt: up to 5,000 characters.

Open it in [Video](https://sloosh.ai/studio/video) and pick Grok Imagine in the model chip. The Generate button shows the cost first.

## Frequently asked questions

### How many reference images does Grok Imagine take?

Grok Imagine takes up to 7 reference images in Sloosh, the same number xAI documents. Say in the prompt what to keep from them, for example "Keep the sneaker's colours exactly as in the reference."

### Can Grok Imagine use a start frame and an end frame?

Grok Imagine takes an optional start frame but no end frame, and in Sloosh a start frame can't be combined with reference images. xAI's model rejects a last frame, and a first frame used together with references.

### Which version of Grok Imagine does Sloosh use?

Sloosh's Grok Imagine calls xAI's classic `grok-imagine-video` model, not Grok Imagine Video 1.5, which xAI released on 31 July 2026 with better motion, physics and audio.

### How do I write a Grok Imagine prompt?

Open with the subject and action in the order they happen, name the camera move, say what to keep from the reference, and end with a "Sound:" line that describes sounds by material and place. Keep to one clear action per clip; several actions can glitch.

## Sources

- [xAI: reference to video](https://docs.x.ai/developers/model-capabilities/video/reference-to-video)
- [xAI: video generation](https://docs.x.ai/developers/model-capabilities/video/generation)