Grok Build pricing
Grok Build is available to try free. The official individual-plan page lists SuperGrok at $30/month and SuperGrok Plus at $100/month. Using an API key adds a different budget: charges depend on the model and tokens used, rather than a fixed monthly coding fee.
Find the right billing route →Account plans and API usage cover different things
| Route | Published price | What to check |
|---|---|---|
| Free account | $0/month | Build trial access and current account limits |
| SuperGrok | $30/month | Higher usage limits across the account's features |
| SuperGrok Plus | $100/month | Higher usage across Chat, Imagine, Voice and Build |
| xAI API key | Metered tokens and applicable tools | Exact model, endpoint, context band and service tier |
The monthly amounts above are the USD individual-plan cards shown on the official pricing page on the verification date. Check checkout for your billing region, taxes and billing period. They are account subscriptions, not standalone Build API packages. The public cards do not promise a fixed number of coding tasks or API tokens, so there is no reliable dollars-per-task conversion.
Check how the coding agent is authenticated
The Build setup guide describes browser authentication on first launch and an API-key option for environments without a browser. It also supports custom model configurations. Before estimating a session, check the active model and authentication route. A screenshot of a plan price tells you little about a script running against an API key.
For account access, start with the plan and usage information shown in your signed-in product. For an API integration, review usage and billing in the xAI Console. Do not assume an existing subscription provides an unlimited API balance. If a plan or promotion explicitly grants credits, budget from those terms and the balance displayed for that account.
Which model does Grok Build use?
The current Grok Build product page says it is powered by Grok 4.7. Some illustrated terminal sessions still display 4.6. Check the model selected in your own session before borrowing a price from an older tutorial.
There is also an API model named grok-build-0.1. Its official aliases include grok-code-fast-1, grok-code-fast and grok-code-fast-1-0825. That older model name does not identify the current default of the Build application. Our Grok Code Fast 1 API price page keeps the existing URL and explains the alias mapping.
A reproducible API budget
Suppose an integration explicitly calls grok-build-0.1 for 1,000 requests in a month. Each request contains 20,000 fresh input tokens and produces 2,000 billable output tokens. At Standard rates, 20M input costs $20 and 2M output costs $4, for a $24 token bill. This is an illustrative workload, not a measured coding project or a subscription allowance.
Open this $24 workload in the calculator. The existing calculator supports this model and preselects it. It does not estimate a Grok Build subscription or automatically switch the estimate to Grok 4.7.
If 80% of those input tokens are reported as cache hits, the same workload costs $4 for fresh input, $3.20 for cached input and $4 for output: $11.20. Cache hits must actually occur. Retrying work, expanding prompts or generating more output changes the bill, even if the visible user task stays the same.
Keep the 200K boundary in the estimate
The official Build 0.1 price table charges $1 input, $0.20 cached input and $2 output per million tokens below 200,000 prompt tokens. At 200,000 or more, those rates double for the entire request. Its context window is 256,000 tokens; monthly aggregate usage is not the context size of one request.
For example, one request with 200,000 fresh input tokens and 1,000 output tokens costs $0.404. The higher input rate applies to all 200,000 input tokens, rather than just the portion near the threshold. Both calculators on this site apply the inclusive boundary for this model.
The API pricing reference lists Grok 4.7 separately: below 200K prompt tokens, its Standard input/cache/output rates are $2/$0.50/$6 per million; at 200K or more, $4/$1/$12. These are public API list prices. They are not an invoice for an account-authenticated Build session.
Fast, Priority and Batch need separate checks
According to the Grok 4.7 developer guide, 4.7 Fast is available through Cursor and Grok Build plans, is excluded from Build's free tier, and is not available on the public API. Its plan billing should not be confused with API Priority processing.
API Priority carries a 2× token premium when the response confirms the Priority service tier. Build 0.1 does not support Batch, and the estimates here assume no Flex discount. Paid tools, regional endpoint premiums, retries and taxes are excluded from the $24 example. A tool-heavy agent can therefore spend more than the simple token calculation.
Choose a budget you can verify
For occasional interactive work, check the free account's current Build access before paying for higher limits. For automated jobs, record the provider endpoint, exact model ID, fresh and cached input, output, and processing tier. Reconcile those fields with actual usage after a small representative run.
The xAI price catalog helps compare listed models. Use the preselected Build 0.1 calculator when that is the model your integration calls. Token-rate differences alone do not establish which agent finishes your project for less money.