Grok 4 adds a text-to-video feature called “Imagine”, powered by Aurora AI

A new creative tool will let Super Grok users generate short videos from text prompts, including audio and spicy content options.

Grok will launch a multimodal video feature starting in October 2025

Grok is about to take a major step toward becoming a full-scale generative multimedia platform. Elon Musk's AI division, xAI, has officially confirmed that a new tool called “Imagine” will launch in October 2025, initially reserved for Super Grok subscribers. The feature allows users to input short textual prompts and receive back six-second videos with synchronized audio, created entirely by Grok’s proprietary Aurora AI engine.

This move places Grok in direct competition with upcoming offerings from OpenAI’s Sora and Google’s Veo, though Musk’s version will likely distinguish itself with user-centric access models, X platform integration, and provocative content settings that are already drawing attention. The rollout is expected to be phased: early access will be granted via waitlist only through the Grok standalone app.

“Imagine” turns prompts into audiovisual clips through the Aurora engine

The videos generated by Grok’s “Imagine” feature are not simple animations or GIFs. They are full audio-visual sequences—with speech, sound effects, and visual transitions—produced in response to plain natural language prompts. According to xAI engineers, the Aurora AI model is capable of handling not only realistic textures and motion synthesis but also voice inflection, timing, and ambient audio layering.

Early examples suggest that prompts such as “a cat playing piano on the moon” or “a classroom turning into a jungle” produce surreal but narratively coherent micro-videos, similar in spirit to shortform Vine clips. Each prompt results in multiple video variants, giving users the option to select, download, or share their favorite output. The final result is typically capped at 6 seconds, suggesting a strategic design for shareability on X (formerly Twitter), where Grok is natively integrated.

Content modes include “Normal”, “Fun”, and a controversial “Spicy” setting

One of the most discussed aspects of the feature is its content tiering system, which includes at least three visible settings: Normal, Fun, and Spicy. The last option—marked explicitly as allowing adult, NSFW, or erotic content—has already sparked debate online. The system appears to use age restrictions and opt-in toggles, but critics warn that this level of creative freedom could open the door to misuse, especially in the era of deepfakes and intimate image abuse.

Musk’s team defends the inclusion on the grounds of artistic freedom, stating that users should be allowed to “create anything within the law.” However, advocacy groups like the National Center on Sexual Exploitation have already flagged the tool as a potential vector for non-consensual deepfake porn, urging regulators to impose stricter guardrails before public release.

The launch aligns Grok with a new category of creative AI interfaces

The broader implication of “Imagine” is Grok’s formal entry into the multimodal AI creation arena, where the boundaries between text, image, video, and audio are no longer separate interfaces but components of a single creative process. Unlike OpenAI, which has split these features across products (e.g. Sora for video, DALL·E for images), Grok now places all modes—including voice generation and image rendering—under one unified toolset, accessible within the same prompt window.

This also reflects Musk’s vision of AI as an integrated assistant for both productivity and entertainment, where creative ideation, content generation, and social publishing are compressed into one interface. “Imagine” is not just a toy or a gimmick—it’s a real-time production engine, one designed for both virality and storytelling in the modern content economy.

A high-stakes rollout exclusive to Super Grok users—for now

The first version of “Imagine” will be exclusive to Super Grok subscribers, the top-tier plan priced around $30/month, which also includes early access to Grok Agent and deep reasoning tools. Users will need to download the separate Grok app and sign up via a waitlist system, as slots are expected to be limited during the beta testing phase.

No formal date has been announced for a full public release, and there are rumors of regional restrictions, especially in countries with strict content regulations. However, Elon Musk hinted on X that the video generation capabilities will eventually be expanded to include longer durations, custom voices, and music scoring features—positioning Grok as both a storytelling tool and a possible disruptor in short-form content platforms like TikTok and Reels.

Technical snapshot of Grok's “Imagine” video feature

Attribute	Details
Name	Imagine
Model	Aurora AI (proprietary text-to-video + audio generation model)
Availability	October 2025 (early access via waitlist in Grok app)
Length per video	~6 seconds
Modes	Normal, Fun, Spicy
Output formats	MP4 (standard), with downloadable/shareable links
Subscription tier	Super Grok (~$30/month)
Content filters	Age-gating and NSFW toggles for spicy content
Core use cases	Social media, storytelling, marketing, satire, NSFW entertainment
Future roadmap	Longer videos, custom voices, musical accompaniment, region unlocks

_________

FOLLOW US FOR MORE.

DATA STUDIOS

datastudios.org