Grok Video API Guide: Features, Pricing & Code

Grok Imagine, xAI's image and video generation system, works in two ways: an image mode and a video mode built for quick, social-ready clips rather than high-production realism. This guide covers the video side specifically: what it does, what it costs, how to access it, and how to use it in your code.

What Is Grok Imagine's Video Mode

Grok Imagine's video capability turns a text prompt, or a reference image, into a short video clip. It sits in the same space as Veo 3, Sora 2, and Kling, but it's positioned differently: xAI has leaned into speed and iteration over maximum realism, which makes it a common pick for teams that want to generate a lot of variations quickly rather than one polished final result.

xAI shipped the Grok Imagine API to developers in early 2026, and followed it with Grok Imagine Video 1.5 mid-year, which added stronger motion and physics along with support for combining multiple clips into a longer video. If you used the earlier version and found motion a bit stiff, 1.5 is worth trying.

It's worth being clear about scope here: this piece is about Grok's video generation, not the separate Grok Imagine image mode, which has its own pricing and its own comparison articles.

Key Features

Prompt-to-Video Generation

Send a text prompt and get a video clip back. Grok's style tends to lean punchy and stylized rather than photorealistic, which fits well with meme content, social-first shorts, and quick concept work.

Image-to-Video Mode

You can pass up to seven reference images alongside a prompt, and Grok will use them to guide the generation, which is useful for keeping a consistent character, product, or scene across a clip.

Duration, Resolution and Audio Support

Clips run 6, 10, or 15 seconds depending on what you request, with resolution generally capped around 720p on the standard video model (Grok Imagine Video 1.5 supports higher resolution on text- and image-to-video, though reference-to-video generation stays at 720p). Unlike Veo 3, Grok doesn't generate audio alongside the clip, so if a clip needs sound, you're adding it in a separate step.

Generation Speed

Speed is one of Grok's selling points. It's generally faster to generate a clip through Grok Imagine than through Veo 3 or Sora 2, which matters if your workflow depends on quick iteration rather than a small number of final-quality outputs.

Content Policy Notes

Like every major video model, Grok Imagine has content moderation on what it will generate, and the specifics of that policy have been a point of public discussion since launch. If your use case is anywhere near a sensitive category, it's worth testing prompts against the current policy before building a production workflow around it, since moderation rules get adjusted over time.

Grok Video API Pricing

xAI's pay-as-you-go developer API bills per second of output, and the exact rate has shifted a few times since the API's early 2026 launch as new versions have shipped. Public reporting through mid-2026 put direct video generation somewhere in the $0.05 to $0.08 per second range depending on resolution and version, with 720p output priced higher than 480p. Given how often these rates have moved, the most reliable number is whatever's listed on xAI's current API pricing page at the time you're budgeting a project, rather than any figure quoted in an older article, including this guide.

On the consumer side, xAI also sells Grok Imagine access through its SuperGrok subscription tiers (a lower-cost SuperGrok Lite tier around $10 a month with reduced resolution and clip length, and the full SuperGrok tier around $30 a month), but that's separate from the pay-as-you-go API and worth mentioning mainly so you don't confuse the two when comparing costs.

Aggregator pricing (through platforms like Apiframe) is billed in credits rather than a direct per-second dollar rate, with cost scaling by clip duration. It's usually worth comparing both routes against your expected volume before picking one.

How to Access the Grok Video API

xAI publishes its API documentation and access details through its developer console. Getting a key means signing up for an xAI developer account and enabling billing, similar to other pay-as-you-go AI generation tools. Access has broadened significantly since the initial 2026 rollout, though it's still worth double-checking current account and region requirements before you build against it, since availability details can change.

The alternative is going through an aggregator. Apiframe (https://apiframe.ai) exposes Grok Imagine Video through the same unified endpoint it uses for every other video model, which means you get one API key, one billing relationship, and the ability to route a request to Grok, Veo 3, Kling, or Sora 2 by changing a single model parameter, rather than setting up each one separately.

Code Example

Here's a Grok Imagine Video generation request through Apiframe's video API (https://apiframe.ai/docs/videos/grok-imagine-video):

import requests

response = requests.post(
    "https://api.apiframe.ai/v2/videos/generate",
    headers={
        "X-API-Key": "afk_your_api_key_here",
        "Content-Type": "application/json",
    },
    json={
        "prompt": "a futuristic city skyline at sunset with flying cars",
        "model": "grok-imagine-video",
        "grokImagineVideoParams": {
            "duration": 10,
            "aspect_ratio": "16:9",
        },
    },
)
data = response.json()
print(data)  # {"jobId": "...", "status": "QUEUED"}

Like every video model on Apiframe, this returns a jobId immediately and runs in the background. Poll GET /v2/jobs/{job_id} until status reads COMPLETED, or add a webhookUrl to receive a notification when it's done. The grokImagineVideoParams.duration field accepts 6, 10, or 15 seconds, and aspect_ratio supports "1:1", "16:9", "9:16", "4:3", "3:2", or "2:3".

For image-to-video, pass reference images instead of relying on text alone:

response = requests.post(
    "https://api.apiframe.ai/v2/videos/generate",
    headers={
        "X-API-Key": "afk_your_api_key_here",
        "Content-Type": "application/json",
    },
    json={
        "prompt": "animate this scene with gentle camera movement",
        "model": "grok-imagine-video",
        "grokImagineVideoParams": {
            "duration": 6,
            "reference_images": ["https://example.com/scene.jpg"],
        },
    },
)

reference_images accepts up to 7 image URLs, which get used to guide the generation toward a consistent look.

Grok Video vs. Other Video Models

Grok's main advantage is speed and low cost for high-volume, fast-iteration workflows, not native audio or maximum photorealism. If your project needs synced dialogue or ambient sound baked into the clip, Veo 3 or Sora 2 are the better choice. If you need to generate a lot of quick variations for social content and can add sound separately, Grok is worth testing first. For more video model options, check our guides on Pika Labs and PixVerse.

Use Cases

Grok Imagine Video shows up most often in social-first short video content, where speed and volume matter more than maximum fidelity. It's also popular for rapid prototyping and creative ideation, letting teams test a concept in video form before committing to a more expensive generation pass on another model. Meme and reactive content is a natural fit given the platform's roots, and the image-to-video mode gets used regularly for animating marketing stills into short motion assets.