Grok (xAI)

What it actually does

Grok is xAI's assistant, reachable inside X and as a standalone app. The differentiator is real-time access to X data, so it can report what is trending and what people are posting about a topic right now. The other major assistants cannot do that. It also handles vision, code generation, image generation and search, with a humor mode that loosens the tone.

It is less filtered than ChatGPT or Claude and will answer prompts they decline. Consumer sentiment is enormous and positive: the US App Store listing shows 4.9 from around 1.4 million ratings and a number 8 spot in Productivity, with users citing real-time X and web access, image and video generation, and voice mode. Against that, it is weaker than GPT-4o on complex tasks, tied to the X ecosystem, and thinner on features than ChatGPT.

What it costs

There is a free tier on X with limited queries. Standalone access is $8 a month. X Premium+ at $16 a month includes it.

At $8 it is the cheapest paid assistant we cover, and that is the honest appeal. Two things temper it. App Store review text surfaced in search described the app hitting rate limits extremely quickly, though we saw that in a search listing rather than verified on the page, so treat it as unconfirmed. And the $16 route only makes sense if you want X Premium+ anyway, since the assistant alone is $8.

Where it falls down

The dominant issue is safety behaviour, and it is a matter of public record rather than user grumbling. Grok's image tools were used to produce sexualised images of real people, including minors, and regulators responded.

The European Commission opened a formal Digital Services Act investigation on January 26, 2026 into whether X assessed and mitigated the risks arising from Grok's functionalities, after Grok-generated sexualised images of women and minors spread on X. DSA breaches carry fines of up to 6% of global annual turnover. Reuters reported that the Commission had said earlier that month that "the AI-generated images of undressed women and children being shared on X were unlawful and appalling," and quoted EU tech chief Henna Virkkunen saying "Non-consensual sexual deepfakes of women and children are a violent, unacceptable form of degradation."

Ofcom opened its own investigation into X over the same imagery roughly two weeks earlier, reported January 12, 2026. Indonesia, the Philippines and Malaysia temporarily blocked Grok; the Philippines and Malaysia restored access after xAI said it had added safety measures. On January 14, 2026 xAI said it had restricted Grok image editing and blocked location-based generation of images of people in revealing clothing, without naming which jurisdictions were affected.

Both investigations are open. Nothing has been decided against X or xAI, and we are not going to guess how they end. What is settled is the sequence: the capability shipped, it was used at scale to produce that material, and the restrictions came after the backlash rather than before the launch.

On quality, one App Store review excerpt described Grok as "the weak link right now" next to other assistants, again seen in a search listing rather than verified on the page.

Who should buy it

Heavy X users and people tracking breaking news on that platform get something here that no rival offers, and $8 is a low bar to clear.

Businesses should not buy it. If you are expensing an AI assistant, expect to explain the choice, and two live regulator investigations into sexualised imagery are a hard thing to explain to a client, a board or a compliance officer. The reputational exposure runs to the buyer, not just the vendor.

Do not buy it as your main assistant on capability grounds either. It sits behind GPT-4o on complex work, and ChatGPT or Claude will do more for $20. Do not buy it for image generation at all, given the feature was restricted mid-flight and the restrictions are geofenced in ways xAI has not specified.

If you want the real-time social feed and nothing else, the free tier on X gives you that without a card on file.