Guides

Grok Pricing 2026: Free, SuperGrok, and API Costs

#Grok#xAI#Grok 4

What Grok's Free, SuperGrok, and SuperGrok Plus plans cost

xAI's pricing page confirms three plans with a published dollar figure: Free at $0/month, SuperGrok at $30/month, and SuperGrok Plus at $100/month. Free is pitched as a way to try Grok within usage limits. SuperGrok adds higher rate limits and access to frontier models, and SuperGrok Plus goes further with heavier usage, 1080p video creation, and priority access at peak times.

The free plan also carries its own usage cap that xAI keeps separate from the weekly limit paid plans share. The Grok FAQ does not publish a specific request count or token figure for that free-tier cap.

You will still have access to Grok's free tier limits on Chat and Voice
View official source →

What SuperGrok Lite and SuperGrok Heavy add, and X Premium+

xAI's comparison table also lists SuperGrok Lite, SuperGrok Heavy, Business, and Enterprise, but the pricing page shows no dollar figure for them. The tier names appear only as column headers in the plan comparison.

Third-party roundups publish prices for these tiers, but none of those figures appears on xAI's own pages. What xAI's page does show is their order: Lite sits below standard SuperGrok, and Heavy sits above SuperGrok Plus for the heaviest individual use.

X Premium+ works differently: it bundles Grok into an X subscription instead of selling it as a SuperGrok tier. If a specific Lite or Heavy price matters for your decision, check x.ai/pricing or grok.com/plans directly before subscribing.

xAI API pricing for Grok 4.7 and other current models

The xAI developer API is billed separately from the consumer plans, at a rate per million tokens that depends on the model. There is no subscription fee for API access, and every request is metered.

  • Grok 4.7, the current flagship, and Grok 4.6 charge the same rate: $2 for every million input tokens and $6 for every million output tokens, on a 500k-token window.
  • Grok 4.5 shares that same input/output rate and 500k window, but its cached-input rate is lower, at $0.30 per million versus $0.50 for 4.7 and 4.6.
  • Grok 4.3 and the three Grok 4.20 variants (reasoning, non-reasoning, multi-agent) cost less, at $1.25 input and $2.50 output per million, and hold a larger 1M-token window.
  • Grok Build 0.1, xAI's coding model, is the cheapest of the group at $1.00 input and $2.00 output per million, with a 256k-token window.

All of those short-context rates double once a request's prompt reaches 200k tokens. Requests routed through the US regional endpoint carry a further 10% premium on top of whichever rate applies.

Requests sent to the US regional endpoint run inference in the United States; their token usage is billed at 1.1x the global token rates, a 10% premium.
View official source →

How long-context rates and Grok Imagine pricing work

Once a request's prompt reaches 200k tokens, xAI bills the entire request at the long-context rate, not just the portion past the threshold. A prompt at 199,999 tokens is billed at the short rate in full, while one token more moves the whole request into the long-context column. This matters most on grok-4.3 and the grok-4.20 variants, since their 1M-token windows make it easy to cross 200k with a single long document.

Image generation through the Grok Imagine API is priced separately from the text models above, by resolution and quality tier. The 2.0 tier (grok-imagine-image-2.0) runs $0.01 to $0.08 per image, the base tier runs $0.002 to $0.02 per image, and the quality tier (grok-imagine-image-quality) runs $0.01 to $0.06 per image.

What changed since Grok 4 retired, and what 4.3 and 4.20 offer now

On May 15, 2026, xAI retired eight models, including grok-4-0709, grok-3, grok-code-fast-1, and the four "fast" reasoning/non-reasoning variants, plus grok-imagine-image-pro. xAI's migration notice asked developers to pick a replacement for each retiring workload before the cutoff. Reasoning workloads on the retired models moved to grok-4.3 and its successors, with reasoning-effort settings replacing the separate fast and non-fast model names.

Coding workloads moved to grok-build-0.1, and image generation moved to grok-imagine-image-quality. If your own code still pins one of the eight retired model names, requests to it need to be redirected explicitly.

Grok 4.3 itself is unaffected by the retirement: it shipped in April 2026 with a context window of up to 1 million tokens, and Grok 4.20 followed as a flagship beta in three variants, each also listed at up to 1 million tokens. xAI's model page now recommends the newer grok-4.7 as the most capable model for most use cases, including code and chat. That leaves Grok 4.3 and 4.20 in the middle of the lineup, still supported and still billable.

we recommend explicitly choosing the right replacement model for each workload before May 15.
View official source →

Grok vs ChatGPT pricing, and which one costs less

Neither xAI nor OpenAI publishes a page that compares the two products' pricing side by side, so there is no official basis for calling one cheaper. The confirmed figures are Grok Free at $0/month and SuperGrok at $30/month.

Grok's main difference is that it can reference real-time posts on X (formerly Twitter) instead of stopping at a fixed training-data cutoff. Whether it "costs less" for you depends on whether you need that real-time access.

Grok Business, Enterprise, and X Premium+ access

xAI's plan comparison table lists Business and Enterprise next to the individual tiers, but neither shows a list price. For custom limits and volume pricing, xAI directs organizations to its sales team.

X Premium+ is a separate product from SuperGrok: it is an X subscription tier that bundles Grok access alongside X's own perks, rather than a plan sold directly on grok.com. Its price is set on X's own subscription page, not on xAI's pricing page.

Talk to our sales team about custom rate limits, dedicated infrastructure, SSO, compliance requirements, and volume pricing for your organization.
View official source →

Summary: three published Grok plan prices, metered API rates

The only consumer prices xAI's own pricing page confirms right now are Free ($0/month), SuperGrok ($30/month), and SuperGrok Plus ($100/month). SuperGrok Lite, SuperGrok Heavy, Business, and Enterprise all exist as named tiers on the same page, but xAI does not publish their dollar amounts there.

API pricing is metered per million tokens and differs by model: grok-4.7 runs $2.00 input and $6.00 output for short-context requests, doubling once a prompt crosses 200k tokens, plus a 10% premium on the US regional endpoint. On May 15, 2026, xAI retired eight older models, moving reasoning workloads to Grok 4.3 and its successors and coding workloads to grok-build-0.1.

FAQ

Is Grok free to use?
Yes. xAI's Free plan is $0 per month, with its own usage limits on Chat and Voice that are separate from the weekly usage limit paid plans share. Paid plans start at SuperGrok for $30 per month.
What does SuperGrok Plus cost, and what does it add over SuperGrok?
SuperGrok Plus is $100 per month, confirmed on x.ai/pricing. It includes everything in the $30 SuperGrok plan plus 1080p video creation, significantly higher usage across Chat, Imagine, Voice, and Build, and priority access at peak times.
How much does the xAI API cost per model?
Rates are quoted per million tokens and vary by model, for example Grok 4.7 at $2.00 input and $6.00 output. Prompts of 200k tokens or more are billed in full at the higher long-context rate, and the US regional endpoint adds a 10% premium.
What happened to the older Grok models?
xAI retired eight models on May 15, 2026, including grok-4-0709, grok-3, and grok-code-fast-1. Reasoning workloads moved to Grok 4.3 and its successors, and coding workloads moved to grok-build-0.1.