Grok Subscription vs xAI API Pricing: Which Route Pays for What?

Since October 12, 2026, Grok is sold through The X Subscription: Free at $0, Plus at $8, Premium at $30, Super at $100 and Ultra at $200 a month, each paid tier with one weekly pool shared across Grok, Grok Bot and Cursor. Plus and higher include Grok API access, but API usage is billed separately at standard model rates and never draws from that pool, so the two routes share no balance. This page explains which route pays; use Grok pricing for the maintained purchase matrix.

Choose the ledger before the plan

Actor, credential, and billing owner are more reliable than a shared product name. An account session or supported organization login points toward a Grok product allowance. A direct API key points toward the API team's ledger. Grok Build can sit on either side, so its resolved authentication method matters more than the word Build.

The X Subscription, top-ups and older plans

Free is the no-cost starting route. Plus at $8 adds the base weekly pool, Grok API access billed separately from it, and X creator benefits. Premium at $30 adds Grok Bot, video generation and four times the Plus pool, and Super and Ultra raise the pool to 15 and 50 times Plus. Every paid tier shares one weekly pool across Grok, Grok Bot and Cursor, actions draw different amounts, and the usage view shows the next reset date. There is no honest universal conversion from a monthly fee to a message quota, and this guide does not invent one. Grok usage limits by plan explains the pool on each plan.

The Grok API line on the plan card does not merge the ledgers. SpaceXAI's pricing FAQ says paid plans include Grok API access but bill API usage separately at standard model rates, outside the plan's usage pool, and its API accounts FAQ says API billing is managed in xAI Console. A Premium or Ultra seat therefore adds nothing to an API budget: forecast every software-driven workload on Console credit at the published token rates.

Top-up credit is the subscription's overflow. It pays for first-party usage beyond the weekly pool and for third-party models in Cursor, and a monthly spending cap can stop paid usage even when credit remains. It does not become reusable credit in xAI Console, and Console prepaid credit does not refill the weekly pool.

X benefits now come with the plan once a personal X account is connected: a checkmark and creator tools on Plus, full X Premium+ on Premium and above. Existing SuperGrok and X Premium subscriptions continue until the subscriber switches or cancels; on SuperGrok, Extra Usage Credits remain the overflow. Switching cancels the old plan with a prorated refund and cannot be undone; The X Subscription explained covers that decision. None of these routes creates an xAI API balance.

What Grok 4.7 API billing funds

The direct API route begins in xAI Console. A developer creates or selects an API team, adds prepaid credit or approved invoiced capacity, creates an API key, and sends requests from software. The team owns the usage record and invoice, so a personal or workspace Grok purchase should not be assigned as the cost owner for a production service.

Grok 4.7 has two published token bands. The long-context threshold is based on prompt length, and once the prompt reaches the threshold the long-context prices apply to all tokens in that request.

A request can create more than one kind of charge. Model input, cached input, reasoning, and output tokens remain billable, while autonomous server-side tool use can add several invocation charges during one user task. Web Search, X Search, and Code Execution therefore belong in the workload estimate even when the base model price looks predictable.

Long context and Priority Processing are separate decisions. Long-context pricing is triggered by prompt length. Priority Processing is requested through the priority service tier and is billed at two times the applicable standard token rates only when the response confirms that tier. Cache discounts are applied before the priority multiplier. Do not merge either condition into the short-context base row.

The API pricing page lists Grok 4.7 Fast separately: 2x the standard token rates below 200K prompt tokens and 1.5x the long-context rates above ($6 input, $1.50 cached input and $18 output per 1 million tokens), sold only inside Cursor and Grok Build and billed through those plans rather than the public xAI API. The fast label and the API documentation's Priority Processing label should not be silently treated as interchangeable, and neither should overwrite the standard Grok 4.7 price. Check the selected model, requested service tier, returned service tier, and per-request cost before assigning the premium.

Use Grok 4.7 vs Grok 4.6 for the generational model-selection boundary it documents. Use Grok web search vs X search when the remaining question is which search tool the workload needs. For current API estimates, start with Grok 4.7 and the official pricing page.

xAI exposes per-request cost data, and Usage Explorer can group consumption by key, model, token type, or billing item. Track prompt length, cache hits, output length, retries, reasoning, tool loops, media units, and request volume. Prepaid credit supplies a natural stop when invoiced billing is disabled; approved monthly invoicing and its spending limit are separate team controls.

Grok Build crosses both routes

Grok Build is the clearest example of why the credential decides the bill. On first interactive launch it can open a browser for account authentication. Enterprise deployments can also use device or organization identity. Those routes use the account or workspace relationship and belong on the Grok product side of the decision.

Grok Build can instead resolve an API key, especially for scripts, CI, bots, and other headless automation. That path calls the direct API and should be forecast from the selected model, prompt and output tokens, tool calls, processing tier, and API-team controls. A Teams or Enterprise login existing in the same organization does not make API-key requests part of a seat.

Current Grok Build documentation identifies Grok 4.7 as the model that also runs directly through the API. That shared model does not merge the commercial routes. Verify the resolved credential before attributing the cost, and treat short-lived included-usage promotions as route-specific rather than reusable API credit.

Teams, Enterprise, and mixed organizations

Teams is the managed people-and-workspace route. Organizations buy Premium, Super or Ultra seats at $30, $100 or $200 per user a month, mix tiers, and get SSO, admin controls, centralized billing and usage analytics. Each seat keeps its own weekly pool, the workspace shares one prepaid balance, and third-party model requests in Cursor carry a $0.25 per million-token Cursor Token Rate on team plans. It fits organizations that need governed interactive use rather than a collection of personal subscriptions.

Enterprise remains a sales-led route with one usage pool across Grok and Cursor and custom organization, security, identity, support, and deployment terms. Because its public price is not a numeric self-serve rate, budget it as a contract discussion. Do not turn Enterprise into a guessed per-seat number or assume the contract includes a general API allowance unless the order form says so.

An xAI API team is a developer billing and credential boundary. It owns API keys, prepaid balances, invoiced usage, usage exploration, and API invoices. Teams and Enterprise may share administrators with an API team, but their seats and the API team's meter remain separate unless a reviewed contract explicitly changes that boundary.

Many organizations therefore need both: seats for people who research, write, analyze files, create media, or work interactively in Grok Build, plus an API budget for customer features, scheduled jobs, internal automations, CI agents, and services that run without a person initiating each request. Track the two routes as different line items with different owners and stop controls.

Final decision check

Before paying, write down who initiates the work, which credential authenticates it, which account or team owns the charge, which unit drains, and which control stops overspend. If an estimate mixes a weekly pool with API tokens, treats top-up credit as API prepaid credit, or assumes a team seat pays for API keys, the routes are still being conflated.

Start with Free when the job is exploratory personal use. Choose Plus, Premium, Super or Ultra when one person needs more Grok product capacity, starting at Premium if Grok Bot or video is required. Choose Teams or Enterprise when people need managed organization access.

Choose the xAI API for software-driven Grok 4.7 workloads, API keys, automation, and per-request cost tracking. Buy both workspace access and API capacity when people need Grok and systems need xAI models. Finish by checking Grok pricing for maintained plan and route facts; the durable rule is that product access and direct API usage have separate billing owners.