Pricing and plans

Grok's pricing is unusual because the consumer product is bundled into an existing social network. That makes it cheap for light users who already pay for X, and confusing for everyone else. This lesson separates the three routes so you can pick one on purpose.

Route 1: the free allowance inside X

Anyone with an X account can open Grok and ask a limited number of questions per time window. The allowance is enough for casual use: summarise a thread, check a fact, draft a reply. It is not enough for daily deep work, and the model you get may be a lighter one.

If you find yourself waiting for the limit to reset more than once or twice a week, you are a candidate for a paid plan. If you never hit it, stay free. There is no prize for paying.

Route 2: X Premium tiers and SuperGrok

Paying for X Premium unlocks higher Grok limits and stronger models inside the X app. SuperGrok is the standalone Grok subscription for people who want the assistant without the social-network bundle. Both are flat monthly fees; the exact amounts and what each tier includes are on the official xAI page, so see current pricing there before deciding.

A flat plan is worth it when you use Grok most days and you value not thinking about tokens. It is a poor deal if you use it in bursts, because you pay for the quiet weeks too. Our subscription vs API guide walks through the break-even logic in detail.

Route 3: the xAI API

The API is billed per token, per model. At the time of writing, Grok 4 Fast lists at $0.20 per 1M input and $0.50 per 1M output; the mid-tier Grok 4.3 and Grok 4.20 list at $1.25 and $2.50. A typical chat message of a few hundred tokens costs a fraction of a cent on any of them.

xAI also publishes a cheaper rate for cached input tokens, which applies when you resend the same long prefix, and a batch rate on some models for jobs that can wait. We do not quote the cached figure here; see the official pricing page. Lesson 7 shows how to take advantage of both.