Skip to content
Back to all articles
By TJ SoftUpdated:

Free AI Videos with Sound in Grok (and Bulk Graphic Creation). [Update]: As of March 2026, Available Only on the Paid Plan.

Filmy AI za darmo i z dźwiękiem w Groku (oraz masowe tworzenie grafik). [Aktualizacja]: od Marca 2026 tylko w płatnym planie.

Grok, the AI from X (Twitter), has a very useful, though little-known, feature. It is Grok Imagine, a tool for generating videos with sound and creating multiple images at once. Technically, these are not new developments: VEO 3 has enabled the convenient creation of such videos for over a year, while Sora (from OpenAI) raised the bar even higher (and other options existed earlier for more technically minded users).

The quality of videos and images created by Grok is nothing groundbreaking. But its advantages are its low cost and ease of use. The entire service is available for free (free is a fair price).

Update [March 2026]. Currently, the imagine features, i.e. bulk image generation, and video creation are available only on the paid plan (approx. PLN 150 per month). You can also access Grok’s paid features by purchasing a premium account on Twitter (X), which costs 8 dollars (approx. PLN 30).

Table of contents

.

How does image generation work in Grok?

The basic thing is to go to the Grok.com website or download the app. Grok available on Twitter (X) has limited features.

Next, you need to go in and select the option: Imagine.

Imagine

We enter the command at the bottom, and that’s it:

Imagine Commands

Within seconds, the program generates eight very high-quality images, and as you scroll, it creates more. This speed made me stop using ChatGPT, where even in the paid version, you have to wait several or several dozen seconds for a single image. However, I have not given up on Google’s Nano Banana Generator, which works instantly and creates images almost immediately. Although in Gemini we get only one image instead of a dozen or so, the editing quality in Nano Banana means that I still use this generator.

As you scroll, new graphics are smoothly generated.

We can enter text in Polish, and the model understands it.

The results are rather random because Pachycephalosaurus is not a very popular dinosaur. In one of the images, Grok even mistook it for Pac-Man (because PAC hycephalosaurus). The correct dinosaur appears on average in one out of every four images; for example, in the picture to the right of Pac-Man, we can see a human figure with a Pachycephalosaurus head. In the case of Velociraptor, the results are much better. For example, Google’s VEO 3 had a problem with this just after its launch and “generated generic” dinosaurs.

Video Generation:

Simply click the arrow icon in the bottom-right corner of the image to have Grok generate a video. The tool allows you to create multiple pieces of content simultaneously, including in the background. Clicking on the image also gives you the option to describe exactly what should happen in the video.

. .

Additionally, by clicking the photo settings (the three dots in the upper-right corner), we can access the editing options. This feature can be useful, although it does not always work satisfactorily. For example, attempting to give the dinosaur a realistic appearance did not produce good results, but moving it to a city worked without any problems. When it comes to photo editing, Nano Banana Pro remains the undisputed king.

.

One trick is to use video mode to modify a photo. This is particularly useful when changing a pose or movement, as we can select a specific frame, although content changes also work. In the video description, simply enter, for example: “add a red hat to the dinosaur” or “the dinosaur puts on a red hat.”

Option: upscale video: allows you to improve the video quality. It is useful if you are starting with a low-quality image.

.

We can create multiple versions of the same video, always starting from the base image. However, if we want to create a complex story, we should copy the last frame (by right-clicking—there is no need to take screenshots) and then paste it as the beginning of a new thread.

."""

What are the limits for generating images and videos in Grok Imagine?

In Imagine mode, Grok allows you to generate between 60 and 180 images per day. This is usually enough for a dozen or so prompts, but the limit can be reached very quickly if you scroll through the results intensively. After exceeding it, images can still be generated directly in the chat window. For videos, the limit is much lower—Grok allows you to create only a dozen or so video clips per day.

Pasting photos:

If we paste an image into Grok Imagine mode, the program immediately creates a video based on it by default. It happens so quickly that there is no time to enter a prompt describing the scene. In such a case, it is much more convenient to paste the image directly into the chat.

.

After pasting a photo into the Chat, we can select the editing option:

And we still have the option of creating a video; this time, we have more time to describe the planned result.

.

Of course, you can also create images in a regular thread without entering image editing mode.

A language model writing promotional copy based on an image.

Sometimes, good results can be achieved by asking Grok to prepare a prompt that will produce a similar image. This technique can also be used in ChatGPT or Gemini, but in Grok’s case, thanks to its mass image generation capability, it is particularly useful. To conserve the limit in Imagine mode, it is worth testing prompts directly in the chat thread. In this case, only two images are generated at a time, and they do not count toward the pool of images created in bulk.

. .

Copyrights

When uploading photos and transforming them into videos, you must of course remember about copyright. Using someone else’s work as a base (the so-called source image) to generate a video with AI may infringe the author’s economic rights to the original. This applies particularly when the video clearly reproduces the unique features, characters, or composition of a photo for which we do not hold the rights. It is safer to use our own photos or graphics generated by us from scratch, although even then the legal status of a work created by artificial intelligence remains unclear in many countries.

In short, the problem may arise from using AI-generated graphics as elements of brand identity, such as a logo or a banner on a website or social media. In short, we may have a problem with the legal protection of such graphics, meaning that potentially anyone may use them. Unless we can demonstrate a significant contribution of our own to generating such images. Illustrative graphics are not a major problem; sometimes the fact that someone copies our graphics does not pose a significant threat to the brand.

Changes to Grok Imagine over the past few months:

For some time, the quality of images created directly in Grok threads was significantly lower than those generated in Imagine mode. Today, these differences have largely disappeared. Grok uses two groups of image-generation models: its proprietary Aurora model and the external Flux model from Black Forest Labs. Previously, the chat window primarily used Aurora, while Imagine relied on the better Flux model. Today, the quality is comparable, although it is still difficult to determine conclusively what criteria Grok uses to choose a particular generator.

Advantage and drawback of Grok: much less censorship than in other models.

ChatGPT and Google Gemini (Nano Banana) have very strict restrictions. These models refuse to create images depicting well-known people, even with completely safe prompts. Grok has no such problems. The situation is similar with adult content: ChatGPT and Gemini block harmless commands. Grok has some restrictions (it does not allow nudity to be generated directly), but skimpy clothing is not an obstacle for it. What is more, with a little creativity, these restrictions can be bypassed. Sometimes Grok even adds nudity to images on its own, despite the prompt not suggesting anything of the sort and not being an attempt to bypass the filters. I suspect this is one reason for the oversensitivity of censorship in other models: an image generator may add various “artifacts” that are then blocked by safety filters.

For adults, the lack of censorship is not a major problem; the problem arises when children use this tool. Therefore, when creating courses on using AI for children and young people, it is better not to promote Grok.

Summary:

Grok Imagine is a powerful tool that lets you create videos with sound completely free of charge. For AI-generated animations, the quality is satisfactory. The video feature can also be used to edit photos—I have used it myself many times when Google Nano Banana couldn’t handle the task. Another major advantage is the generous free limits and the ability to create graphics quickly and in bulk.