Is Grok Imagine Free? Images Yes, Video No (2026)

Quick answer: for still images, yes. xAI’s own plan comparison table shows image generation checked under the Free column as of September 10, 2026, and video generation dashed. The most widely cited paywall story is a January 2026 restriction on X itself, and the pages still carrying it have not looked at the pricing page in months.

My read on Grok Imagine: the stills are the free part, the video is the paywall, and nobody selling you an alternative wants you to know the difference.

Published September 10, 2026. Plan rows verified on xAI’s pricing page the same day.

Quick decision

Your situation	What to do
You want still images and you pay nothing	Use it; xAI’s table checks Imagine on Free
You want video	That is the paid line, on every tier above Free
You want 1080p video	That is the $100 tier, SuperGrok Plus
You want to know your limit	Nobody publishes one. Here is the meter
You are choosing a rung to pay for	Free vs SuperGrok, per trigger

xAI’s own table says images are on the free plan

This is the whole page, so I am putting the receipt first.

xAI’s pricing page, captured September 10, 2026. Image generation (Imagine): checked under Free. Video generation: dashed under Free, checked on everything above it.

Two rows, one conclusion. Stills are in. Motion is what you are actually buying.

That single dash is worth more than the ten blog posts above this one on Google, because it tells you exactly which half of the product is behind the wall.

The widely cited January 2026 story happened on X

So why does the entire first page of search results tell you Grok Imagine is not free?

In January 2026, X restricted Grok’s image generation and editing for non-paying users on X itself, after a backlash over what people were making with it. Gulf News covered it at the time and noted something the aggregators dropped: people who were not paying could still reach Grok’s image features through the separate app and website.

That nuance did not survive. “You cannot edit an image in an X reply anymore” became “Grok Imagine is paid,” and the pages carrying that claim have been ranking ever since without anyone rechecking it.

Look at who is ranking for this question and you will see why nobody corrected it. Most of page one is either a post from before the change or a site selling you a different image generator, with the disappointment as the hook.

xAI’s own pricing page contradicts itself, and you should know that

I am not going to pretend the answer is cleaner than it is.

Same page, same day, scrolled up. The Free card lists four things. Images is not one of them. “Image and video generation” first appears on the $30 card.

So the comparison table says images are on Free and the plan card does not mention them at all. Both are on xAI’s pricing page right now.

My read, and I want to be clear that it is a read: the table is the more specific artifact because it distinguishes stills from video row by row, while the card is marketing shorthand for what the $30 unlocks. The card is not wrong that images get better above Free. It just is not the place to learn whether the free plan can make one.

Here is what I checked and what I did not. I read xAI’s published table on the day of writing, on my own paid account, and I screenshotted it. I did not create a free account, so I have not personally watched a $0 login press generate and get a picture back. If you are on free right now, you can settle it in ten seconds, and your ten seconds beats my inference.

What I would not do is trust a number. xAI publishes no free-tier image cap, so every “3 images a day” and “5 to 15 a day” figure you will find is someone’s estimate rather than something xAI has stated. Paid plans have run on one shared weekly compute pool since June 2026.

And here is the third contradiction, which cuts against me rather than for me. xAI’s FAQ says the free tier keeps its own limits on “Chat and Voice” that reset on their own schedule. It names Chat and Voice. It does not name Imagine. If you wanted to argue the free plan does not really get images, that sentence is your best evidence, and I would rather hand it to you than pretend the table is the last word.

What an hour on the paid version actually produced

I have a SuperGrok Plus account, so this section is the paid experience. I am not going to dress it up as a test of the free tier.

I typed one thing and got four.

One prompt, four variations, with Generate More and Think Harder waiting underneath. The play button on each tile is the video path.

Four variations from one prompt is the part that changes how you work. You are not iterating on one image and hoping. You are picking.

There is also a Discover surface, which is where you go to find out what the thing is capable of before you have any idea what to ask it for.

Discover, same session. “Generate a cartoon” with no other direction.

It will not make you copyrighted characters. Ask for Mario and it refuses, which is the correct answer and also not a real limitation for anyone making a header image.

I will say the thing I did not expect to say: I used to think this was trash. It is not trash.

One is hands-free, the other is a control surface

This is the distinction that actually decides which tab you open, and I have not seen anyone else write it down.

	ChatGPT images	Grok Imagine
How you drive it	Describe it, it decides	Toggle settings, dimensions, elements
Closest comparison	An assistant	Midjourney
Good when	You want it handled	You know what you want and want to steer
The cost of that	Less control	More decisions

With OpenAI you say it and it generates. It is more included, more hands-free, and that is a genuinely good experience when you do not want to think about it.

Grok Imagine goes the other way. You can select different elements, change specific things, and get very specific about dimensions. It almost feels like a Photoshop.

Neither of those is a criticism. They are two different products for two different moods, and knowing which mood you are in is the entire decision.

Not everyone agrees with me, and here is who

I formed my impression in about an hour. Named people who spent longer landed in different places, and you should have both.

TechRadar tested the image editing features and published the verdict in the headline: they are fun but will not replace Photoshop any time soon. That is a direct contradiction of my “almost feels like a Photoshop” reaction, and I am leaving it in rather than picking the quote that agrees with me.

We are describing different things. I meant the interaction model feels like an editor rather than a slot machine. They mean it will not do your actual retouching work, and they are right about that.

Notebookcheck filed it under “where AI brilliance meets mild trauma,” which is about as accurate a summary of using this product as three words can be.

invideo put it as comparable to mid-tier specialist models on stills and trailing the premium options on motion. That matches the pricing table, which is a decent sign both are describing reality.

What flipped me, specifically

It was not one output. It was that it was intuitive, and that it was fast.

The generation speed is the thing. It felt slightly faster than the OpenAI image model I normally use, and I did not time it, so take that as an impression rather than a benchmark.

What I can count is the output: 4 options where I expected 1. That is the part that changes behavior, because you experiment instead of committing.

On raw quality, I am not going to overclaim. Compared against the newest OpenAI image model, which gets genuinely precise on the fine detail like skin and wrinkles, Grok is not the precision winner. But it generates some flawless images, and the gap is much smaller than it was the last time I looked.

That is the honest shape of it: I went from dismissing it to keeping it open in a tab.

Does it change my stack? Partly, and I will say where it does not

Yes, more than I expected, but not everywhere.

• It adds a real option for images. I thought the field was Nano Banana and OpenAI’s image model. Grok has earned a place in that set.
• It will not make my thumbnails. No reason to switch. What I use works, and thumbnails are the one place where I am not experimenting.
• It might make blog images. A quick explanatory graphic, the kind of thing that is not worth a long session. That is the use case I would test next.
• I will not be running it through a CLI. For me this is a browser tool, not part of an automated pipeline.

If I am honest, I have not found the killer use case for it in my own work yet. It is on the radar now, which it was not last week, and that is a real change even though it is not a migration.

Where this fits

If your question is really about which plan to pay for, that is free versus SuperGrok, one trigger per rung, and the $100 question is answered separately. If you got here from the app-building side of Grok, Build mode is the other thing that is quietly on the free plan. For how this sits against ChatGPT generally, Grok vs ChatGPT.

The rest of what I actually run is on my Claude at work hub.

How I checked this: I read and screenshotted xAI’s pricing page on September 10, 2026, and used Grok Imagine on my own SuperGrok Plus account the same day. I did not test a free account, and I say so above rather than implying otherwise. Outside verdicts are linked to the publications that wrote them. Last reviewed September 2026.