Grok NSFW and Grok Spicy Mode Limits xAI No Longer Publishes

What is Grok spicy mode?

Grok spicy mode is the informal name for an adult content setting inside Grok Imagine, the part of Grok that generates images and video. Turning it on widens what the generator will produce, in the direction of suggestive rather than explicit. It is a media setting. It is not a switch that makes the text assistant talk freely about anything you raise.

The confusion starts because a second Grok feature, Companions, also gets called spicy mode in casual writing. Companions are persistent character personalities you hold conversations with, and they are the thing most people actually mean when they say they want an uncensored Grok. Those are two different products inside one app, with different rules and, critically, different device support. Companions are also being retired: the Grok App Store listing stopped mentioning them with the version released September 30, 2026.

If you take one thing from this page, take that separation. A guide that tells you to verify your age and toggle a setting is describing Imagine. A guide that describes a character who remembers you is describing Companions. Following the first set of instructions will never produce the second result, which is why so many people report that the steps did not work for them.

• Grok Imagine: image and video generation, has an NSFW setting.
• Companions: character conversation, iOS app only per xAI documentation.
• Grok Chat: the general assistant, not marketed as an adult product at all.

Does Grok spicy mode still work?

Possibly, but xAI no longer says so anywhere. Until the end of August 2026 its help FAQ carried an entry for people who enabled NSFW and found content still being blocked, and we used to point to it as the best evidence the setting was alive, because a company does not keep a troubleshooting answer for a feature it removed. Between August 31 and September 3, 2026 that entry was deleted, and nothing replaced it. The feature may still be in the apps. The only written acknowledgement of it is not.

If it is not working for you, check three ordinary causes before concluding it was removed. Paid plans share a weekly usage pool, and paid features pause when it runs out. A subscription only shows on the login that bought it. And moderation was documented as staying on even with NSFW enabled, so a refusal is not a removal. What has genuinely changed is that you can no longer check any of this against an xAI page about the adult setting, because there is none.

Is Grok NSFW?

Partly, and only in a specific place. Grok is the only assistant from a major AI lab that ships a sanctioned adult setting at all, which genuinely distinguishes it from OpenAI, Anthropic and Google. Give it that credit. But the setting governs generated media inside Imagine rather than the assistant as a whole, and it operates under moderation that stays on.

For a buyer the practical question is not whether a product is NSFW in the abstract. It is whether the specific thing you want to do will be refused. Grok will refuse a large amount of adult text conversation even with every toggle enabled, because the toggle was never aimed at that. That is not a criticism of the product. It is a mismatch between what it is for and what people are searching for when they type these words.

Does Grok spicy mode mean it becomes 18+?

Enabling it signals that you are an adult and unlocks adult-leaning output, but it does not convert Grok into an adults-only service with adults-only rules. The floor stays where it is. xAI's Acceptable Use Policy names the permanent exclusions directly, including sexualizing or exploiting children and altering a real person's image to depict them in an intimate or sexual context, and until September 2026 its FAQ added that those protections cannot be disabled by any setting or subscription.

It is worth being clear that we hold the same lines, and we would not take a customer who wanted them removed. Nothing involving minors, nothing involving non-consent, nothing illegal. Where we differ from Grok is not the floor. It is everything above the floor, where Grok applies an unpublished layer of additional moderation and we do not.

Why is Grok spicy mode gone for some people?

Three ordinary reasons account for most of it, and none of them require a conspiracy. The first is surface. Features in this space do not ship everywhere at once, and xAI has documented at least one case, Companions, where a surface gap is permanent by choice rather than temporary by schedule. The second is the usage pool. Paid Grok plans now share one weekly allowance across Chat, Imagine, Voice and Build, and xAI states that when the weekly limit is met, paid features pause until it resets. A paused feature looks identical to a removed feature.

The third is moderation doing its job. Because xAI will not publish which prompts are blocked, a refusal carries no explanation, and a user who is refused three times in a row reasonably concludes the feature was taken away. It usually was not. It just declined, silently, for a reason nobody will tell you.

That opacity is the real cost, and it is worth weighing before you subscribe to anything in this category. A product that refuses you without saying why is a product you cannot plan around.

Can you use Grok spicy mode on the web?

For Companions, no, and this is settled rather than pending. xAI states in its FAQ that Companions are available on the iOS app only and that there are no plans to bring them to the web or Android. That single sentence answers a question a large number of people are searching, and it answers it in the least convenient direction. Since late September 2026 the question is moot on iPhone too, because the App Store listing no longer advertises Companions at all.

Think about what that means if you own an Android phone or work on a laptop. The character conversation that every Grok roundup describes is not delayed for you, and it is not in beta for you. It is not coming. No settings path, no browser extension and no account tier changes that, because it is a product decision the vendor has written down.

This is the specific gap we were built into. Depravity runs in a browser, so the same account behaves identically on an Android phone, an iPhone, a Windows laptop and a tablet. There is nothing to install and no store that has to approve us first.

For what the Companions retirement means if you paid for Ani, see <a href="/lp/grok-ani-nsfw">Grok Ani NSFW and the alternatives</a>.

Is Grok Imagine spicy mode free or paid?

Paid, according to every guide, and xAI will not say which plan. The prices themselves are now readable at the source: on September 21, 2026 the x.ai pricing page listed SuperGrok at $30 a month and SuperGrok Plus at $100, and the App Store purchase list added SuperGrok Lite at $10, a $300 SuperGrok item that appears to be the annual plan, and SuperGrok Heavy at $300 with no period stated. Guides say spicy mode needs SuperGrok at $30 and that Lite is not enough. None of the plan descriptions xAI publishes mentions spicy mode, NSFW or Companions, so that is a claim you are buying on trust.

What we can tell you is the shape of the cost, because xAI documents that. Paid plans get a weekly usage pool spent across every Grok product, and the company notes that different products consume it at very different rates: a chat message uses little compute, while generating high-quality video uses far more. If you generate a lot of media, you can exhaust the same allowance your conversations draw on. Extra Usage Credits expire a year after purchase; the help FAQ says they can only be bought on the web, while the App Store lists them as in-app purchases at $5, $20, $50 and $100.

Every plan, its price and what xAI does and does not say it unlocks is on our SuperGrok NSFW and spicy mode price page.

What does Grok spicy mode still refuse?

The permanent exclusions are stated: anything sexualizing minors and non-consensual intimate imagery, neither of which any legitimate service will ever allow, ours included. Above that floor sits a layer xAI described until September 2026 but never detailed, and that layer is where most real refusals happen.

The company frames the secrecy as an anti-abuse measure, and that reasoning is defensible. Publishing an exact list of blocked prompts is publishing a bypass guide. The tradeoff is that paying customers cannot tell in advance whether the thing they signed up to do is permitted, and are warned that finding out by trying can put the account at risk.

Our approach is the opposite trade. We publish our three hard lines and stop there, which means the boundary is legible before you pay. We also concede what that costs us: no image generation, no video, no public character catalogue and no first-party mobile app. Grok does more things than we do. It just does not do this one thing on your Android phone or in your browser.