Grok Imagine: Image and Video From xAI, Explained

Grok Imagine is xAI's image and video generator, built into the Grok app and also sold through xAI's API. It makes pictures from a prompt, turns a photo into a short clip, and — on the video side — adds sound to the clip automatically instead of leaving it silent.

What Grok Imagine actually is

"Grok Imagine" isn't one model, it's a product name covering two separate engines: an image model and a video model, both from xAI. Grok Imagine creates images and short videos from text prompts and reference images, edits visuals with instructions, and generates videos with native audio. The current API generation is Grok Imagine Image 2.0 for stills and Grok Imagine Video 1.5 for clips.

The headline feature is speed, not cinematic polish — xAI has leaned into fast turnaround over photorealistic physics, which shapes what the model is good for.

Grok Imagine Image

The image side takes a text prompt (and optionally a reference image) and returns a picture. You can generate up to 10 images per request at 1K or 2K resolution across a wide range of aspect ratios. It also edits: feed it an existing image with an instruction and it will change just the part you describe.

Grok Imagine Video

The video side does image-to-video and text-to-video, and its point of difference is sound. Instead of a silent clip you then score yourself, Grok Imagine Video 1.5 renders dialogue-free audio (ambience, effects, sometimes voice) in the same pass as the picture. In the consumer Grok app, video generation up to 15 seconds at 720p is included on both the free and paid tiers.

What it costs, straight from xAI

If you go through xAI itself, there are two ways to pay: a flat subscription, or metered API pricing.

The consumer route is SuperGrok. SuperGrok costs $30/month and includes extended access to Grok 4 along with use of the Imagine model, a more capable image generation tool. The free tier of the Grok app also gets image and limited video generation, just with lower daily limits.

The API route is metered per output, and the rates are specific to model and quality tier:

Source: grok-imagine-image at $0.002/img input and $0.02/img output (1K or 2K); grok-imagine-image-quality at $0.01/img input and $0.05/img (1K) or $0.07/img (2K) output; grok-imagine-video at $0.01/sec plus $0.002/img input and $0.05/sec (480p) or $0.07/sec (720p) output; and grok-imagine-video-1.5-preview at $0.01/img input and $0.08/sec (480p) or $0.14/sec (720p) output.

xAI's own docs summarize it more simply: image generation starts at $0.02 per image at 1K or 2K resolution, and video starts at $0.05 per second at 480p, 720p, or 1080p. The gap between that headline number and the 1.5-preview table above is the newer, higher-quality model costing more per second — worth checking before you batch a long run.

How it compares to other video models

Grok Imagine isn't trying to win on realism. Sora 2 Pro is the benchmark for cinematic video realism, with detailed motion and physically consistent environments, while Grok Imagine trades that quality ceiling for speed and versatility, with faster generations and an integrated image-to-video pipeline at a lower cost per output. That's the trade-off to keep in mind: pick Grok Imagine when you need a lot of quick variations or built-in audio, and pick a model built for realism when a single shot needs to hold up under scrutiny.

For stills, Grok Imagine 2 sits in the middle of the pack on price and leans toward sharp typography and clean edits rather than the heaviest photorealism — for that, Nano Banana 2 or GPT Image 2.5 go further, covered in our best AI image generator comparison.

Do it in Plurel

You don't need a SuperGrok subscription or an xAI API key to use either model — Plurel runs them on the same credit balance as every other model in the app. Sign up free with Google, no card required.

To make an image with Grok Imagine 2:

1. Open https://plurel.tech/?model=grok-imagine-image-2, or click Image under the composer and pick Grok Imagine 2 from the model list.
2. Type your prompt — include style, subject, and any text you want rendered in the image.
3. Generate. Each image costs 2 credits.
4. To change it, just follow up in the same chat ("make the background darker", "add a logo in the corner") — that's an edit on the picture you just made, not a new one from scratch.

To animate a photo with Grok Imagine Video 1.5:

1. Open https://plurel.tech/?model=grok-imagine-video-1-5, or use the Animate tool at https://plurel.tech/?model=animate and select the model.
2. Upload the photo you want to bring to life.
3. Describe the motion and the sound — Grok Imagine Video 1.5 renders audio in the same pass, so mention it directly: "waves crashing, seagulls in the distance."
4. Generate. Pricing starts at 18 credits, roughly half a credit per second of output, so longer clips cost more.
5. To keep going, type "continue that clip" in the same chat — Extend picks up where the last one ended.

If a chat model is driving the shot for you, Grok 4.7 can plan the sequence and hand it to these tools directly — open it at https://plurel.tech/?model=grok-4-7 and ask it to draft a prompt before you approve the credit cost.

Sample prompts

A chrome robot barista pouring coffee in a sunlit café, 2K, soft morning light, shallow depth of field

Image prompt for Grok Imagine 2 — tests detail and light handling.

Neon sign reading "OPEN LATE" above a rain-soaked noodle shop doorway, reflections on wet pavement

Image prompt for Grok Imagine 2 — tests text rendering inside the scene.

Animate this photo: the subject turns their head slowly toward the camera and smiles, soft wind moving their hair, ambient street noise

Image-to-video prompt for Grok Imagine Video 1.5 — tests native audio and subtle motion.

A paper boat drifting down a rain gutter, close-up, water splashing, light rain sound, 8 seconds

Video prompt for Grok Imagine Video 1.5 — tests audio sync on a small, controlled scene.

Product shot of a matte black water bottle rotating slowly on a marble surface, studio lighting, 1K

Image prompt for Grok Imagine 2 — tests clean product-style renders.

A street musician playing guitar, crowd noise and a faint melody, camera slowly pushes in

Video prompt for Grok Imagine Video 1.5 — tests layered ambient sound with a simple camera move.

Tips and troubleshooting

If a video comes out with no sound or mismatched sound, describe the audio explicitly in the prompt rather than assuming it will guess — "ambient café chatter" works better than nothing at all. If an image looks over-smoothed, add a lighting and lens detail ("35mm, soft side light") rather than piling on adjectives. And if a clip is too short for what you need, don't re-prompt from scratch — use Extend or just ask the chat to continue it, which is cheaper than regenerating the whole shot.

Grok Imagine's edge is turnaround and built-in sound, not maximum realism — pick it for fast iteration, and reach for a heavier model when a single frame has to carry the whole piece.