How to use Grok 4.7 for free?

The only confirmed free way to run Grok 4.7 is Grok Build's free tier. Every other route checked: Grok app, Cursor, Copilot, xAI API, Vercel, OpenRouter.

Ashley Innocent

Ashley Innocent

29 September 2026

How to use Grok 4.7 for free?

Apidog for Enterprise

On-Premises Deploy

SSO & RBAC

SOC 2 Compliant

Explore Apidog Enterprise

The one confirmed way to run Grok 4.7 without paying is Grok Build, SpaceXAI’s coding agent, which uses Grok 4.7 as its default model and has a free tier. SpaceXAI doesn’t publish that tier’s limits. The other popular routes don’t get you free Grok 4.7: the Grok chat app’s plans are still marketed on Grok 4.6, the xAI API needs prepaid credits, and Vercel, OpenRouter, and GitHub Copilot Free all leave it out of their free allowances.

That’s a shorter list than most “free Grok” pages promise, so this guide separates what’s confirmed from what isn’t, flags the offers to skip, and shows how to make a small paid budget go a long way by mocking the API in Apidog during development. For the model itself, see what Grok 4.7 is. For free Grok in general, including the chat app and Imagine, our Grok free guide covers the wider picture.

Every route, checked

Route Free? Runs Grok 4.7? What to know
Grok Build free tier Yes Yes, the default model Limits not published; Grok 4.7 Fast excluded
Grok chat app (grok.com, apps, X) Yes Not confirmed Plans page says “Chat with Grok 4.6”; free limits not published
Cursor Hobby Yes Unconfirmed Cursor’s docs list Grok 4.7 on individual and team plans
GitHub Copilot Free Yes No Grok 4.7 is on Pro, Pro+, Max, Business, and Enterprise only
xAI API No Yes Prepaid credits; no sign-up credit in the docs
Vercel AI Gateway $5 every 30 days Not with the free credit Grok 4.7’s row is marked “Free Tier: No”
OpenRouter 50 requests/day on free models No x-ai/grok-4.7 has no free variant

Route 1: Grok Build’s free tier (confirmed)

Grok Build is SpaceXAI’s terminal coding agent. Its product page says it’s “now powered by Grok 4.7” and available to try for free, and the Grok 4.7 developer docs name Grok 4.7 as its default model. The only exclusion the docs call out is Grok 4.7 Fast, the 2x-priced faster variant, which isn’t in the free tier.

Setup takes a few commands on macOS or Linux:

curl -fsSL https://x.ai/cli/install.sh | bash
cd your-project
grok

On Windows, the installer is irm https://x.ai/cli/install.ps1 | iex. On first launch, Grok opens a browser to sign you in. You can also run it headless for one-off questions:

grok -p "Explain the architecture of this repo"

What you don’t get is a published quota. SpaceXAI doesn’t say how many requests or tokens the free tier includes, or whether it ever asks for a card. Expect a cap and plan around it:

One caution: Grok Build is an agent that edits files and runs commands. Use it on a branch and review diffs before you commit.

Route 2: the Grok chat app (free, but probably not 4.7)

Grok is free to start on grok.com, in the mobile apps, and inside X. The Grok help FAQ confirms free users get their own Chat and Voice limits, separate from the paid weekly pool, without publishing numbers.

The catch for this guide: as of September 28, 2026, the Grok plans page headline reads “Unlock the full power of Chat with Grok 4.6.” No official page says free users, or even paid chat users, can pick Grok 4.7. The free chat tier is a fine way to use Grok, but not a reliable way to use Grok 4.7. Check the model name in the picker before assuming.

The paid chat tiers are SuperGrok Lite, SuperGrok, SuperGrok Plus, and SuperGrok Heavy, drawing from one weekly usage pool across Chat, Imagine, Voice, and Build. Prices didn’t display on the logged-out plans page, so check them inside your account rather than on third-party sites.

Route 3: Cursor’s Hobby plan (unconfirmed)

Cursor’s free Hobby plan needs no credit card and includes “limited Agent requests.” Whether those requests can use Grok 4.7 is unclear. SpaceXAI’s docs say Grok 4.7 is in Cursor “on all plans,” while Cursor’s own Grok 4.7 model page places it in the Cursor Models pool on individual and team plans, and Cursor’s help pages say Hobby users get a smaller model set.

Open the model picker on a Hobby account. If Grok 4.7 is listed, you have a second free route. If not, Cursor’s paid Individual plan advertises “generous limits for Grok.” Our Cursor pricing guide covers the tiers.

What isn’t free (and the offers to skip)

The xAI API. The quickstart tells you to sign up and load credits. The current docs mention no sign-up credit and no data-sharing credit program, so the “$25 free” and “$150 a month” figures on credit-tracker sites are out of date at best. API credits are non-refundable, so start small.

Vercel AI Gateway. Free Vercel accounts get $5 of gateway credits every 30 days, but the Grok 4.7 provider row is marked “Free Tier: No,” so that credit can’t be spent on spacexai/grok-4.7. Vercel’s 40% launch discount was scheduled to end on September 27.

OpenRouter. The free plan covers 50 requests a day on OpenRouter’s free models, and x-ai/grok-4.7 isn’t one of them.

GitHub Copilot Free. Grok 4.7 is on Copilot Pro, Pro+, Max, Business, and Enterprise. Free and Student users get automatic model selection only, and Grok isn’t in that pool.

“Free Grok 4.7 API key” sites. SpaceXAI doesn’t hand out free keys. Pages offering them, “unlimited free Grok API” wrappers, and discounted shared SuperGrok accounts can’t be verified against SpaceXAI’s own pages, so don’t paste account details into them. Keys come from console.x.ai; our Grok API key guide shows the real process.

The cheapest paid route: a few dollars of API credit

If Grok Build’s free tier runs out, a small API top-up goes further than you’d expect. Grok 4.7 costs $2 per million input tokens and $6 per million output tokens for prompts under 200,000 tokens. A request with 2,000 input tokens and 1,000 output tokens (reasoning included) costs about a cent, so $5 covers roughly 500 calls of that size.

Make it go further:

The entry rate-limit tier, for accounts under $50 of cumulative spend, still allows 150 requests per second, so limits won’t be your constraint on a small budget. The Grok 4.7 API guide has the request code.

Stretch paid credits with Apidog

Most wasted API spend happens during development, not production: re-sending the same request while you fix a UI, or re-running a test suite that hits the live model every time. Apidog cuts that out:

A day of frontend work against a mock costs nothing. The same day against the live API can burn through a free tier or a $5 top-up. Download Apidog to set up the mock.

FAQ

Is Grok 4.7 free? Through Grok Build’s free tier, yes, and that’s the only free route SpaceXAI has confirmed. The API is paid, at $2/$6 per million tokens.

Can free Grok app users use Grok 4.7? Not according to any official page. The Grok plans page markets chat on Grok 4.6 as of September 28, 2026. Check the model picker in your account.

Does xAI give free API credits for Grok 4.7? Not in the current docs. The quickstart requires prepaid credits, and the sign-up and data-sharing credit programs older guides mention aren’t documented anymore.

Is Grok 4.7 free in Cursor or GitHub Copilot? Copilot Free, no. Cursor’s free Hobby plan is unconfirmed; check the model picker.

Were earlier Grok models free the same way? The pattern repeats: new Grok flagships start in paid and developer channels first. Our Grok 4.5 free guide shows how it played out last time.

Start with Grok Build

Install Grok Build, sign in, and give it a real task on a branch. If you outgrow the free tier, a $5 API top-up at medium effort plus a mock for your development loop keeps the cost close to zero. If you’re deciding whether Grok 4.7 is worth paying for at all, the comparison with GPT-6 Sol and Claude Opus 5.5 lines up the prices and benchmarks.

button

Explore more

Claude Haiku 5.5 vs GPT-6 Luna

Claude Haiku 5.5 vs GPT-6 Luna

Haiku 5.5 vs GPT-6 Luna: same $0.10/$0.50 price under 100K tokens, different long-prompt tiers, Anthropic's benchmarks, and per-effort cost data.

8 October 2026

Claude Haiku 5.5 Benchmarks

Claude Haiku 5.5 Benchmarks

Claude Haiku 5.5 benchmarks: 72.4% OSWorld, 1620 GDPval-AA, 39.2% Terminal-Bench at max effort. Who ran each test, per-effort costs, and your own eval.

8 October 2026

Claude Haiku 5.5 vs Haiku 4.5: What Changed and the Breaking Changes to Fix First

Claude Haiku 5.5 vs Haiku 4.5: What Changed and the Breaking Changes to Fix First

Haiku 5.5 vs Haiku 4.5: 90% cheaper up to 100K tokens, 1M context, and five breaking changes that return 400s. Before/after JSON fixes inside.

8 October 2026

Practice API Design-first in Apidog

Discover an easier way to build and use APIs

How to use Grok 4.7 for free?