Quick answer: Grok 4.7 came out on September 21, 2026, at exactly Grok 4.6's price: $2 per million input tokens, $6 per million output. It is a real but small step up. xAI's own coding scores rise 6 to 17 points, and Artificial Analysis moves its overall index just 2.
xAI launched it in Cursor, Grok Build and the API. The launch page never mentions the Grok app or SuperGrok.
My verdict after running it in Grok Build: same price, slightly better. I'm like, cool. But these are minor updates. And the first job I gave it failed.
My read on Grok 4.7: cheap per token, hungry per job, third in agentic coding by its own founder's count.
Published September 22, 2026, the day after launch. Prices, benchmarks and availability were checked on xAI's, GitHub's and Artificial Analysis's own pages that day.
Grok 4.7 at a glance: same price, small step upQuestion
AnswerRelease date
September 21, 2026 (xAI's launch post)Price
$2 in / $6 out per 1M tokens, same as 4.6Context window
500,000 tokens (xAI's docs)Real step up from 4.6?
Small. Coding scores up 6 to 17 points; overall index up 2Where xAI launched it
Cursor, Grok Build, the Grok API, coding tools, model routers, clouds; GitHub Copilot rolling outIn the Grok app / SuperGrok?
Not on xAI's launch list. Signed out, grok.com's picker shows modes, not model numbersGrok 4.7 Fast
Same model, twice the speed, twice the price; Cursor and Grok Build onlyMy verdict
"Same price, slightly better." A minor updatexAI's launch page, September 22, 2026. Last line: "Served at the same price and speed as Grok 4.6."
Grok 4.7 shipped to Build, Cursor and the API first
Here is where xAI says Grok 4.7 runs, word for word from the launch post: "Grok 4.7 is available today in Cursor and Grok Build. It is also available through the Grok API, third-party coding harnesses, and model routers and cloud platforms."
Read that list again. There is no Grok app in it, no grok.com, no SuperGrok, no Heavy.Surface
Has Grok 4.7?
SourceGrok Build (coding agent)
Yes, the default model per xAI's docs; chat-window Build mode showed 4.7 on my account
xAI's launch post and docsCursor
Yes, on all Cursor plans
xAI's docsGrok API
Yes, model id grok-4.7
xAI's docsOpenRouter, Vercel, Cloudflare
Yes
xAI's docsGitHub Copilot
Rolling out gradually, paid plans only
GitHub's changelogGrok app chat (grok.com, SuperGrok, Heavy)
Not listed. The signed-out picker shows no model numbers
xAI's launch post; screenshot belowGrok Bot
Unknown. It will not say which model it runs
Grok Bot itself, see belowgrok.com signed out, September 22, 2026. The picker shows modes, not models. Nothing says 4.6 or 4.7.
GitHub's wording is "Rollout will be gradual," for Copilot Pro, Pro+, Max, Business and Enterprise. So if your Copilot picker does not show it yet, that is expected.
Grok 4.7 Fast is the one to watch on your plan. Per xAI's docs it is "the same model served on faster infrastructure, billed at twice the standard token rates," it lives only in Cursor and Grok Build, and "it is not included in Grok Build's free tier."
My first job for Grok 4.7 in Build failed
I have been running Grok 4.7 in Grok Build. We confirmed the 4.7 label in Build, so that is the one surface where I know which model I am on.
The first thing I tried was having Grok Build make an app, or a script, to help me find good business ideas. Just to test it. It failed.
For what it is worth, this is something Codex flunked too, earlier. So 4.7 did not clear a bar that Codex had already missed.
That is honestly all I have tested it on so far. We have been building, or at least failed to build, and that is the whole field test. There is no log of that run on my machines, so I am not going to describe the error or pretend I timed it.
If Build mode itself is new to you, start with what Grok Build actually is. If you are choosing between it and Claude's version, that is Grok Build vs Claude Code.
Grok 4.7 vs 4.6: coding up 6 to 17 points, overall up 2
My honest reaction to the upgrade: "Same price, slightly better. I'm like cool. But it's just like these are minor updates, you know what I mean?"
The numbers say the same thing, with one surprise in the coding rows.Measure
Grok 4.6
Grok 4.7
ChangePrice per 1M tokens (in / out)
$2 / $6
$2 / $6
NoneCursorBench 4.0 (xAI)
40.4%
46.3%
+5.9 pointsTerminal-Bench 4.0 (xAI)
20.3%
37.6%
+17.3 pointsDeepSWE v1.1 (xAI)
65.2%
71.0%
+5.8 pointsGDPval, office-work Elo (xAI)
1605
1695
+90Artificial Analysis Intelligence Index
44
46
+2Artificial Analysis Coding Agent Index (with Grok Build)
47
56
+9Output tokens per index task (Artificial Analysis)
~36k (high)
~81k (xhigh)
More than doublexAI's numbers are xAI's own runs. The Artificial Analysis write-up is the independent check, and it agrees on the shape: a big jump on coding agents, a nudge on everything else.
Compare that to the same week at Anthropic. Claude Opus 5.5 landed one day later, and Anthropic says it "performs at the level of Claude Fable 5.1 on most work" at $4 / $20 per million tokens. That is the bigger update. My full take is in Opus 5.5 vs Fable.
Grok 4.7 is cheap per token, not cheap per job
When I first talked this through, I called 4.7 "way more token efficient." That is half right, and the half that is wrong matters for your bill.
Per token, it is cheap: $6 per million output tokens, against $20 for Claude Opus 5.5 and $50 for Claude Fable 5.1. Per job, it is hungry. Artificial Analysis measured about 81,000 output tokens per task at xhigh effort, against 36,000 for Grok 4.6 and 27,000 for GPT-6 Astra at max.
So the efficiency play is real. It is just the price tag doing the work, not a model that says less.
A Hacker News user, mchusma, noticed the same thing on day one: "It feels kind of like they really had it burn tokens to claw up the benchmarks."
To be fair to xAI, its own CursorBench chart plots Grok 4.7's average cost per task at roughly half of Fable 5.1's or Opus 5's, which is where the launch tagline "half the price of comparable models" comes from. Cheap tokens can still beat expensive ones even when you burn more of them.
Third place, by Musk's own count
What I heard going in was that 4.7 would be at Opus 5 level, and it turned out worse on some things. That is close to what Elon Musk actually posted a week before launch.Musk's pre-launch post, September 14, 2026: 4.7 "roughly on par with Opus 5.0, not 5.1."
After launch he was blunter. His September 21 post, via the Musk post archive: "Grok 4.7 places @SpaceXAI as third, after Anthropic & OpenAI, for agentic coding. When factoring in that Grok is significantly faster & lower cost, it's a great choice for your everyday workhorse."
Artificial Analysis puts it one spot lower. Grok 4.7 with Grok Build ranks fourth on its Coding Agent Index, behind Claude Fable 5.1, GPT-6 Astra and Claude Opus 5.
One correction to my own memory. I thought Elon said Grok 5 would be at Fable level. His September 14 post actually says "Grok 4.9 is probably Astra/Fable class" and "Grok 5 maybe better than anything." Either way, I am waiting on Grok 5.
Because here is the part I actually care about: if you can get Fable-like intelligence at Grok prices, that becomes very intriguing.
Where Grok genuinely wins: price, speed, and a free Build door
I do not think xAI's models are impressive right now. I also think Grok is not to be messed with at this point. They are playing the efficiency game well. 4.6 was pretty good, and 4.7 is solid, for what it is worth.
Here is what Grok 4.7 has going for it:Price per token. $2 / $6 per million, against $4 / $20 for Claude Opus 5.5 and $10 / $50 for Claude Fable 5.1. The output rate is under a third of Opus 5.5's.
More model, same bill. moojacob on Hacker News: "Apparently Grok 4.7 has 40% more weights than Grok 4.6, but the price ($6 output token, $2 input) is the same." xAI has not published a parameter count, so treat the 40% as reporting, not spec.
Speed. xAI serves it at 4.6's speed, and Grok 4.7 Fast gives you 2x the output speed at 2x the price. Musk's own pitch is "significantly faster & lower cost."
A free door. Grok Build has been on every plan, including Free, since August 19, 2026.Grok Bot will not tell you if it runs 4.7
Grok Bot ironically can't confirm if it's using 4.7, so I can't even tell you.
My guess is that a day after launch, it is. But nobody knows what the Bot uses under the hood; it just does whatever it does. So I am not going to vouch for 4.7 there. Build is the only place I have confirmed it.
If you are weighing the Bot against ChatGPT's agent, that comparison lives in GPT-6 Astra vs Grok Bot.
What to do about Grok 4.7 this week
Nobody should change plans over this release. Here is the move by who you are:If you use free Grok, open Build and try it. Grok Build is on the free plan, and xAI's docs say the Fast variant is held back from Build's free tier.
If you pay for SuperGrok, you already have 4.7 where xAI put it, in Grok Build. In regular chat, xAI has not said, and I have not captured my signed-in picker yet. Build and chat draw from one weekly pool, so read how Grok's usage limits work before you point long Build jobs at it.
If you pay for Claude or ChatGPT, this is not the week to switch. The same week brought Claude Opus 5.5, the bigger update in my book, and GPT-6 Sol. For everyday work, my setup is on Claude guides.
If you code in Cursor, this is where 4.7 makes the most sense: it is on every Cursor plan, cheap per token, and fourth on Artificial Analysis's coding-agent ranking. Try it on a job where speed matters more than the last few points of quality.FAQ
Is Grok 4.7 free?
Partly. Grok Build has been on every Grok plan including Free since August 19, 2026, and xAI's docs say the faster Grok 4.7 Fast variant is not included in Grok Build's free tier. xAI's docs describe the Grok Build coding agent; they do not say which model the chat window's Build mode (the hammer icon) runs on a free account.
Through the API you pay per token: $2 per million input tokens and $6 per million output tokens.
Is Grok 4.7 in SuperGrok?
In Grok Build, yes: xAI launched it there, Build is on every plan, and Build showed 4.7 on my SuperGrok Plus account. For regular Grok chat, xAI has not said. Its launch post never mentions the Grok app, SuperGrok or Heavy, and the signed-out grok.com picker shows modes, not model numbers.
What is the difference between Grok 4.7 and Grok 4.6?
Same price ($2 in, $6 out per million tokens), better coding scores. On xAI's own table CursorBench 4.0 goes from 40.4% to 46.3% and Terminal-Bench 4.0 from 20.3% to 37.6%.
Artificial Analysis moves it only 2 points on its overall Intelligence Index (44 to 46) but 9 points on its Coding Agent Index (47 to 56).
What is Grok 4.7 Fast?
The same Grok 4.7 model served on faster infrastructure, at twice the token price. xAI's docs say it is available only in Cursor and Grok Build, is not included in Grok Build's free tier, and is not on the public xAI API.
Is Grok 4.8 next?
Probably, but nothing is published. On September 14, 2026, Elon Musk posted that Grok 4.8 will be a noticeable improvement and that Grok 4.9 is probably Astra/Fable class. There is no release date, model card, or price for either.
Where this comes from
The facts come from xAI, GitHub, Artificial Analysis and Musk's own posts, all read September 22, 2026. What's mine is my SuperGrok Plus account and one failed Grok Build run on 4.7.
I haven't run my own benchmark, and I don't have a screenshot of my signed-in Grok app yet. Last reviewed September 2026.
Quick answer: if you pressed the hammer icon in Grok's chat window, the comparison you found on Google is about a different product. Build mode is free on every plan, needs nothing installed, and publishes to a live link.
Claude Code is a terminal agent that starts at $20 and works on your real files. Prototype in Build. Move it into Claude Code when it gets serious.
That last line is my whole split on these two: prototype in Build, move it when it gets serious.Update, September 22, 2026: both sides got a new model this week. xAI released Grok 4.7 on September 21 and its launch post lists Grok Build among the places it runs. Anthropic released Opus 5.5 the next day, Claude Code's docs now make it the default on Pro, Max, Team and Enterprise, and Anthropic says "Fast mode for Opus 5.5 is also available in Claude Code."
My split does not change. 5.5 is for serious work. 4.7 is just free work.
Build comes with my SuperGrok Plus plan, and with every other plan too, Free included. So to me it is a free canvas. There's definitely a use case for it, I just haven't found it yet. The full reasoning is further down.You are probably looking at the wrong Grok Build
There are two products with this name, and Google's page one is almost entirely about the one you did not mean.
Build mode, inside the Grok chat window. It lives behind the hammer icon in the composer, alongside the other modes. xAI announced it on July 28, 2026, and their pitch was that there is nothing to install and nothing to configure, and you never have to touch the code unless you want to.
The Grok Build CLI. A terminal coding agent aimed at professional engineering, announced back in May 2026.
Same name. Completely different audiences.
When I pulled the results for "grok build vs claude code" on September 18, 2026, every one that was genuinely about Build was about the CLI. Same on September 22, with three new ones added.
One was about Grok Bot, a third product again. Composio's is the honest one, because it puts CLI in the URL and tells you which thing it reviewed.
So if you came here because of a button you pressed in a chat window, that's a different product. What Grok Build is walks through both products and the mode menu; this page is only about choosing.
The two-products problem got worse with Grok 4.7. When xAI says the new model is "available today in Cursor and Grok Build," its docs mean the coding agent, where 4.7 is now the default model. That is also the Build the outside tests run.
The hammer icon is less clear. xAI's launch post does not say which model Build mode in the chat window runs, and it does not list the Grok app at all.
When I tried Build on my own account, I confirmed it was on 4.7. That is my report from one session, with no screenshot, not something xAI states. My first test there, an app or a script to help me find good business ideas, failed.
The money answer: Build is free, Claude Code starts at $20Grok Build (chat mode)
Claude CodeFree plan
Yes, every plan
NoEntry price
$0
$20/mo (Pro), $17/mo annualInstall anything?
No
Yes, terminal or an IDEWhere it works
Your browser, iOS, Android
The real files and folders on your laptop, plus the saved history of earlier versionsOutput
A live page at a grok.me link
Changes to a real project on your machinePublished limits
None for Build specifically
Plan usage limitsxAI changed the gate and most guides never caught up.The announcement is dated August 19, 2026, and the line that matters is one sentence: "Grok Build is now available on every plan, on the web and on mobile."
The paragraph under it explains why so much of the internet is still wrong: "When we introduced it in July, building in Grok was an Early Beta limited to SuperGrok Heavy."
Three weeks of truth, and then a year of guides repeating it. My own account is the cheapest disproof I can offer: I subscribe at the $100 SuperGrok Plus tier, Heavy mode is not on it, and Build is right there in my composer.
On the other side, Claude's pricing page lists Claude Code under Pro. The Free tier does not have it.
What actually comes out of each one
This is where the two products stop being comparable in a useful way.
Build mode gives you a page with a Publish button. In xAI's words, "Every app you publish gets its own address on grok.me, and you decide who can open it: just you, anyone with the link, or the whole internet." You can point a domain you already own at it instead.
That publish step is the entire reason a non-coder would care, and no benchmark table on page one mentions it.Claude Code gives you changes to something that already exists. It runs in your terminal, on the real files and folders on your laptop, including the saved history of every previous version of your work.
There is no Publish button because publishing is not the job. Modifying a project you already have is the job.
One makes a thing. The other changes your things.
I built a landscaping site in Build mode
On September 10, 2026 I gave Build mode a one-pager for an invented landscaping business and let it run.That is the real preview, with the Publish button sitting in the toolbar. It was a test, not a client job, and nothing got shipped to a real business.
What finished: the page. Hero, sections, copy, all of it live in the conversation, no install, no configuration, nothing to deploy.
It also produced a contact form, which is where a page like this usually falls apart.A form that renders is not a form that delivers mail to a business owner, and that gap is the honest edge of what this tool does for you.
The same build, in sixty seconds:The 60-second version, from my Shorts: Grok builds the app from one sentence and hands you the link.For prototyping, I think it is great. You can generate images in Imagine, generate mockup sites, and get to something you can look at fast.
I do not think "less capable" automatically means bad. Grok is finding a level that is respectable, and the stack you get with the subscription is enticing enough to try.
The moment I move it into Claude Code
Here is the line, and the word that decides it is serious.
Prototype, mockup, something you are building to find out whether it is worth building? Stay in Build. That is what it is good at and it costs nothing.
New, complex, or something other people are going to depend on? That is Claude Code, with Opus 5.5 behind it by default and Fable as the fallback, which is the choice I unpack on Fable vs Opus.
The reason is not loyalty, it is model position, and xAI's own founder is the clearest source on it. He has now said it twice.Before launch, on September 14, 2026, Elon Musk posted: "Grok 4.7 should be roughly on par with Opus 5.0, not 5.1. Better in some ways, worse in others. We need to fix multimodal performance."
Grok 4.7 shipped on September 21. That day he posted again: "Grok 4.7 places @SpaceXAI as third, after Anthropic & OpenAI, for agentic coding. When factoring in that Grok is significantly faster & lower cost, it's a great choice for your everyday workhorse."
xAI's own launch table says the same thing in numbers. On CursorBench 4.0, xAI lists Grok 4.7 at 46.3% and Claude Fable 5.1 at 51.8%.
Then on September 22, Anthropic shipped Opus 5.5 and said it "performs at the level of Claude Fable 5.1 on most work." It is already in my own Claude app.My Claude Code log shows the same switch: at 18:08 UTC that day I saved Opus 5.5 as my default for new sessions, at high effort.
Read what that actually says. Grok's own creator ranks it third and sells it on speed and price.
By Anthropic's account, the model Claude Code now gives you by default sits at the level of the model xAI's own table puts ahead of Grok 4.7. For something that matters, that is not a hard call.
I want to be fair about this, because it would be easy to be snide. I am not saying Grok is bad. I have access to it, I think I should lean on it more for certain builds, and I think it is worth using.
So did the two new models change my split? No. 5.5 is for serious work. 4.7 is just free work.
Grok gets serious for me once it reaches Fable level. I'm hoping that's Grok 5, probably before the end of the year if we're lucky. At that point you're talking about more serious tasks, and you can build really serious stuff.
That timing is my guess, not an xAI promise. Musk's own September 14 post pegs Grok 4.9 as "probably Astra/Fable class."
Right now the cheap, efficient side pays off if you're a developer with an app you just want to maintain. For the everyday user, you're better off using ChatGPT or Claude. This is not where you want to get nerdy. I'm not saying it's a terrible model to talk to, it's actually pretty smart.
For everyday use I still go with ChatGPT and Claude, because right now you're getting way more bang for the buck.
Omid Saffari, one of the few solo operators who has compared all three, lands in the same place, in a review written before Grok 4.7 shipped:For a throwaway side project or to kick the tires, fine. For the refactor your paying customers depend on this week, not yet.Worth noting he is reviewing the Grok Build CLI there, not the chat mode, so take it as a verdict on xAI's coding stack rather than on the hammer icon.
The line is not size. It is consequence.
Would I put it in front of a client
For a prototype, yes, and this is where the two Grok products earn their keep together.
Imagine is where I generate images, thumbnails, and visual mockups of a redesign. Build is where I make the actual working page you can click. For showing a client a direction before anyone has committed a budget, that combination is fast and it is cheap.
What I would not do is hand over a Build-mode page as the finished product for a business that depends on it. Prototype, decision, then build it properly. The prototype is not the deliverable, it is the argument for the deliverable.
One warning before you plan around any of this: xAI publishes no Grok Build usage limit, so every specific number you will find quoted for one was invented by the person quoting it, and Grok usage limits explained tracks what is actually published.
The one published gate so far is about speed. Grok 4.7 Fast, the quicker version billed at twice the token rate, "is not included in Grok Build's free tier," per xAI's docs.
So which one should you openIf you clicked the hammer icon and want to know if it is a toy: it is not, for prototypes. It is free, it publishes, and it is worth an afternoon.
If you have no subscription to either: Build mode is free on every Grok plan and Claude Code is not free at all. Start where it costs nothing.
If you are prototyping for a client: Build for the page, Imagine for the visuals, and do not confuse either with the finished job.
If the thing is new, complex, or something other people will depend on: Claude Code, at $20 minimum, and use the terminal.
If you are choosing a terminal coding agent and you write software for a living: you want the CLI comparison, which is what page one is already giving you. This page is not that.
If you are actually weighing Grok's chatbot against Claude Code: that is a different pairing, covered on Grok Bot vs Claude Code.For the non-Grok side of the same decision, Codex vs Claude Code is the version of this argument where one of the tools is free on every ChatGPT plan. The Claude Guides hub has the rest of the map.
Common questions
Is Grok Build the same as Claude Code?
No. Claude Code and the Grok Build CLI are both terminal coding agents. Build mode in the Grok chat window is a different product that installs nothing and publishes to a link. Most comparisons you will find are about the CLI.
Is Grok Build free?
Build mode is on every plan, per xAI's announcement of August 19, 2026. It was Heavy-only for about three weeks after its July launch, so anything written in July says Heavy only.
Which Grok Build does a "Claude Code vs Grok Build" comparison mean?
Almost always the CLI, the terminal product. If you are comparing Claude Code against the hammer icon in your chat window, most of page one is reviewing something else. The same confusion drives searches for grok build vs codex and grok bot vs grok build, which are three separate xAI products.
When should I move from Grok Build to Claude Code?
When the project becomes something other people depend on. Prototypes and mockups are fine in Build. New, complex work that other people will depend on belongs in Claude Code.
Does Grok Build use Grok 4.7?
The coding agent does. Since the September 21 launch, xAI's docs list Grok 4.7 as its default model, and Grok 4.7 Fast is not on Build's free tier. xAI does not say which model the hammer icon in the chat window runs. The model itself is covered on Grok 4.7 explained.
Which model does Claude Code use now?
Opus 5.5, by default, on Pro, Max, Team and Enterprise. Anthropic says it performs at the level of Fable 5.1 on most work. Neither Opus 5.5 nor Claude Code is on the Free plan.
Correction LogSeptember 18, 2026. Published, three days before Grok 4.7 shipped.
September 22, 2026. Grok 4.7 shipped on September 21 and Opus 5.5 on September 22. Replaced the pre-launch Grok 4.7 answers, added Musk's September 21 post and xAI's own CursorBench numbers, added Opus 5.5 as Claude Code's default, the Grok 4.7 Fast free-tier gate, and a note on which Grok Build xAI means. Noted that Omid Saffari's review predates 4.7. The split itself did not change.Where this comes from
The facts come from xAI's and Anthropic's own posts and docs, read September 22. I pay for SuperGrok Plus and Claude Max, and the app screenshots are mine.
The landscaping build was my own test on September 10. I have not tried the Grok Build CLI, and xAI publishes no Build usage numbers.
Published September 18, 2026. Updated September 22, 2026 for Grok 4.7 and Opus 5.5.
Quick answer: Grok Build is a mode in Grok's chat window. You click the hammer icon under the message box, pick Build, describe what you want, and it builds a working website or app live in the conversation with a Publish button waiting at the end.
There is nothing to install. There is also a completely separate developer product called Grok Build that runs in a terminal, and that is the one Google shows you first.
My rule on Grok Build: prototype in whatever you already pay for, and never let one vendor own all of it.
Published September 10, 2026. Plan availability verified on xAI's pricing page the same day.
Quick decisionWhat you are trying to do
Where to goGet a one-page site or a small app without touching code
Build mode, in the Grok chat windowDay-job software engineering
The Grok Build CLI, or Claude CodeFind out whether your plan includes it
It does. Every plan, including freeWork out what it costs you
It draws on the shared weekly poolDecide whether to pay xAI at all
My $100 SuperGrok Plus verdictYou searched Grok Build and Google handed you a terminal
There are two products named Grok Build, and they are not versions of each other.
The one most search results describe is the CLI, announced May 25, 2026 as "a powerful new coding agent for professional software engineering and complex coding work."
You install it with a curl command. It runs in a terminal, plans before it edits, and works across a codebase. People who have put real hours into it rate it seriously. But if you do not write software, it is not for you.
The one you found in your chat menu is Build mode, announced two months later on July 28, 2026. xAI's own sentence for it: "There's nothing to install and nothing to configure, and you never have to touch the code unless you want to."
Same name. The first one needs a terminal. The second one just needs you to type a sentence.
If you are reading this because you clicked something in Grok and a preview pane opened, you want the second one, and the rest of this page is about that.
Your chat menu has five modes, not four
Every mode guide ranking on Google right now lists four: Auto, Fast, Expert, Heavy. Count the menu.My own session on September 10, 2026. Five modes, Build checked, on a SuperGrok Plus account. The prompt is sitting right there: "build a website for a local landscaping website."
Here is what each one is actually for:Mode
What it does
Gated?Auto
Picks between Fast and Expert for you
NoFast
Answers quickly, thinks less
NoExpert
Thinks longer before answering
Yes. xAI's table shows a dash under FreeBuild
Makes websites, apps, games and dashboards
No. Checked on all seven plan columnsHeavy
Runs a larger group of agents on one problem
Yes, and the most restricted of the fiveHeavy is the confusing one, because Heavy is both a mode in this menu and a plan tier. I am on SuperGrok Plus at $100 and I do not have Heavy mode.
I do have Build. The guides mix those two up when they tell you Build is an advanced feature you have to buy.
xAI does not publish a price for the Heavy tier, so I am not going to print one. All I can tell you is my own account: at $100 a month, I don't get Heavy mode and I do get Build.
One more thing you won't see in the menu: xAI's own FAQ now says "Grok Studio is no longer supported. Use Grok Build instead." If you are searching for Studio, this is where it went.
Build stopped being a top-tier feature on August 19
At launch, Build mode was an Early Beta limited to SuperGrok Heavy subscribers. That was true for three weeks, and it's why half the internet still says you need the top plan.
It stopped being true on August 19, 2026, when xAI published a follow-up post and the line that matters is: "Grok Build is now available on every plan, on the web and on mobile."
An announcement is not the product, so here is the pricing page on the same day.xAI's pricing page, captured September 10, 2026. Read the Grok Build row across all seven columns, then read the Grok Bot and Expert rows for contrast.
You can see the old gate in the reaction to the launch. @FFBuncho posted at Build Mode's announcement that you could publish an app to an xAI domain instantly with no domain contract, and was bummed that it looked like SuperGrok Heavy only for now.
That post was right when it went up. It stopped being true three weeks later, and that's the whole problem with searching this question.
The same August 19 update also added remixing other people's published apps, custom domains and GitHub export. Small sites like Basenor still carry the Heavy-only headline in their URL. That's not a knock on them. It just shows how fast this fact went stale.
What one prompt actually produced in my first hour
I typed one sentence: build a website for a local landscaping website. That is the whole prompt, redundant wording and all.
It read the workspace instructions, connected to a machine, loaded design and scaffold skills, and started building while I watched a preview pane fill in. Then I closed the browser by accident, came back, and it had finished anyway.The finished page. Note the Publish button top-left. The business, the address and the phone number are all invented by the model.
It named the business Rowan Hill, dated it to 1994, placed it in Dutchess County, wrote the headline "Ground that feels as if it has always belonged," and built a working nav that scrolls you down the page.
It also built a real form, not a picture of one, though nothing is wired up to receive what someone types into it.Name, phone, email, a service dropdown and a free-text field, from one sentence of instruction.
My reaction in the moment was "holy shit," and here's why. The output itself isn't that special. What got me was that the input was one careless sentence and the thing finished on its own.
I would not hand that landscaping page to a real business yet
No. It needs tweaking, and the tweaking is my fault, not the model's.
A one-sentence prompt gets you a one-sentence site. Steal the prompt I should have written instead:The brand: the exact colors, the logo file, real project photos.
The place: the town and the service area, named.
The benchmark: one or two competitors ranking first in Google Maps for that keyword in that town, and what those sites actually have.
The brief: what the client wants that the competitors do not do.
Then let it iterate, because the second pass is where it gets good.It is also one page. You scroll, you click the menu bar, it scrolls for you. That's fine for a demo, but it's not a business's website.
So as a demo, yes. As something you invoice for, not off one prompt. What it does change is how cheap it is to show someone a mockup, and that matters when you are pitching local businesses.
If that is the business you are in, the longer version of that pitch is selling AI builds to local businesses.
The other Grok Build is a terminal app for engineers
Briefly, because you may have landed here from a developer comparison.
The CLI is available to SuperGrok and X Premium Plus subscribers, runs interactively or headless, and runs multiple agents in parallel.
Composio put 50-plus hours into comparing it against Claude Code, and Analytics Vidhya ran the same comparison under the headline "I Tested Both So You Don't Have To." I have not run that head-to-head myself, so I am pointing you at people who did instead of faking a test.
I would ignore the SWE-bench numbers floating around this comparison. They come from the vendor and get copied between blogs without credit. And they won't tell you whether a landing page comes out right.
When I wrote this, the model behind the CLI was grok-4.6. Update, September 22, 2026: Grok 4.7 is out, and xAI's launch post puts it in the Grok Build CLI, Cursor and the API.
Grok Build usage limits: the same weekly pool as everything else
There is no separate Build allowance and no published Grok Build weekly limit.
Per xAI's FAQ, Build sits on the one shared weekly usage pool alongside API, Chat, Imagine and Voice, and different products drain it at different rates. xAI publishes no per-tier number for any of them, so anyone quoting you a Build limit is inventing it.
The free tier works differently: it keeps its own Chat and Voice limits that reset on their own schedule, separate from the weekly pool.
I wrote the full mechanics up separately, because it's the thing people get wrong most about this product: how Grok's usage limits actually work.
When to open Build instead of Claude Code
My actual rule, in the words I used out loud: choosing between these is like choosing between Macy's and JCPenney. They are more or less the same thing and it depends what is on sale.
What is on sale, for most people, is the ecosystem they already pay for.If you already pay for X Premium and a SuperGrok plan, Build is a no-brainer. You are not going to go buy a $200 Claude Code plan to do the thing you already have.
If you already pay for Claude Code or Codex, Build is your backup. When I run out of usage credit on Codex or Claude Code, I have Build sitting there for a dummy test or a throwaway prototype.
If you are picking one for serious work, I still think Claude Code and Codex are the two serious players. Lean on Claude Code when the copy has to be right; lean on Codex when the work is research-shaped.
If you are optimizing purely on budget, the interesting case is the top Grok plan. Instead of a $200 Claude Code plan plus a $200 Codex plan, one Grok subscription gets you Build, plus Grok Bot with its own separate usage pool, plus Cursor, which xAI's own Grok Bot launch post names alongside your Grok plan. That one's worth a real look.But here's what I care about more than the feature comparison: you would be living entirely in their ecosystem. That is too risky for me. Do not put all of it on one vendor, however good the bundle looks this month.
The gap between these tools is small enough that people make up personalities for them. What matters more is which company owns all your work if they change the terms.
Where this fits
If you are working out the plan question rather than the tool question, start with what free Grok actually gets you, then whether the paid tier is worth it.
If you came here from the image side of Grok, that is Grok Imagine, which shares this same weekly pool and has a very different answer on what free gets you.
Everything I run day to day lives on my Claude guides hub.
How I checked this: I used Build mode on my own SuperGrok Plus account on September 10, 2026 and screenshotted the session. Plan availability, mode gating and the announcement dates come from x.ai's own pages, read the same day. I have not tested the Grok Build CLI, and I say so where it comes up. Last reviewed September 2026.
Chris Alarcon 
