Grok, Cursor, and X Launch Joint Subscription: Four Tiers Starting at $8

On October 12, a joint subscription bundling the AI assistant Grok, the AI coding tool Cursor, and the social platform X was officially announced. The base package is positioned as free (with usage caps), while the paid portion is divided into four tiers, with monthly fees ranging from $8 to $200, all before tax.

Why three products are packed into one subscription

Grok is the large model assistant from xAI, Cursor is one of the fastest-growing AI code editors in the developer community, and X is Musk's social platform. The three were originally billed separately, and now they're on the same invoice.

From a product logic perspective, this isn't a simple "bundle promotion." Grok provides model capabilities, Cursor provides a landing point for programming scenarios, and X provides distribution and social touchpoints — the subscription model strings together "model invocation — coding tool — daily use" into one chain, so users get all three layers of service in a single payment.

Grok is a large language model assistant launched by xAI at the end of 2023. It was initially distributed as an additional benefit for X Premium+ subscribers and later gradually became independent as a web version and app. Compared with ChatGPT and Claude, Grok's differentiating selling point is its real-time access to X platform's information feed, enabling faster responses to breaking news and trending posts.

Cursor is an AI code editor built on a VS Code fork, developed by the Anysphere team. It integrates completion, inline rewriting, and multi-file agentic programming (Agent mode) into the editor itself, rather than mounting them as plugins. Cursor does not train foundation models itself; instead, it calls APIs from vendors such as OpenAI, Anthropic, and xAI, so "which model to use" is often billed separately in its subscription system.

Understanding these two points is essential to grasp where the three products connect: Cursor is a downstream consumption scenario for model capabilities, X is Grok's distribution channel, and xAI needs to find a high-frequency, must-have entry point for Grok — and programming is precisely one of the most mature paid scenarios for large models today.

Pricing and tiers: what does the $8 to $200 range mean

The announced monthly fee range is $8 to $200 (before tax), with four paid tiers.

Four tiers with a 25x price gap indicate that the feature differences won't be small. The lower-priced tiers likely cover only light usage and basic quotas, while the higher-priced tiers should target heavy developers or teams and may include higher quotas, fuller model permissions, and premium X platform benefits. Specific details about what each tier unlocks are currently quite limited. For individual users, whether it's worth it depends on actual usage — if you only occasionally use Cursor for code completion and occasionally ask Grok questions on X, the lowest tier is probably enough.

The most critical detail: shared weekly quota, not separate allocations

The most easily overlooked and most experience-affecting point in this subscription is that Grok's model invocations in the X app and Cursor share the same weekly quota.

This means the quota you consume chatting with Grok on X and the quota you consume using Grok to write code in Cursor are deducted from the same pool. The invocation intensity in the two scenarios differs enormously — code completion and agentic programming often consume large amounts of tokens at once, while conversations on social platforms are relatively scattered. For users who use Grok in both scenarios simultaneously, this shared quota may run out faster than expected, and the risk of "buying a high-tier plan but getting rate-limited mid-month" is real.

"Quota / rate limit" here refers to the metered upper limit of model invocations, usually measured by token count, number of requests, or a combination of both. Mainstream AI subscriptions commonly use sliding time-window rate limiting, with common periods of 5 hours, 24 hours, or 7 days. A weekly quota is a relatively loose tier, suitable for more fluctuating usage rhythms, but it also means that once you consume intensively early in the cycle, you can only wait for the reset in the later part.

The difference in token consumption across usage modes is on an order-of-magnitude scale. A single round of Q&A on a social platform is usually a few hundred to a few thousand tokens; while a single task in Cursor's Agent mode may read multiple files, repeatedly call tools, and generate and modify code, with single-run consumption of tens of thousands or even over a hundred thousand tokens being not uncommon. Therefore, "sharing one pool" is asymmetrical in practice — the programming side will consume noticeably faster than the chat side. For budget-sensitive users, the actual programming time available may be shorter than the tier names suggest.

Third-party models are not included in the package

Other models in Cursor — such as the Claude and GPT series — are not included in this shared quota system and require separate payment in supported tiers.

This design draws a clear boundary: the joint subscription actually bundles xAI's own model capabilities (plus platform benefits from Cursor and X), rather than "any model in Cursor, use as you like." For developers accustomed to switching between different models in Cursor to compare outputs, the billing structure will be more complex than expected.

Model access in AI coding tools is typically multi-vendor. Cursor allows users to switch between Claude, GPT, Gemini, and other models in settings, passing through costs according to each vendor's API pricing. Such invocations are generally not included in the platform's own subscription quota, but instead use a model of "fast requests" quota plus pay-as-you-go billing beyond that.

This brings up an easily overlooked billing structure issue: the joint subscription covers Grok invocations, but if a developer habitually uses Claude for refactoring, GPT for writing tests, and Grok for quick Q&A, then the costs for the first two will still be incurred separately. In other words, this subscription cannot replace Cursor's original paid plan; it stacks on top of it. When evaluating cost-effectiveness, you need to calculate the "shared weekly quota" and "third-party model pay-as-you-go fees" separately, otherwise it's easy to underestimate actual monthly spending.

What about existing users: can hold off for now

Users who have already subscribed to related services can currently maintain their original plans. Switching to the joint subscription is not mandatory, and the rollout of the new plan is gradual.

This gradual approach gives existing users an observation window: waiting until the quota rules and the actual limits of each tier are verified by early users before deciding whether to migrate is more prudent than switching immediately. Especially regarding the shared weekly quota, it's worth seeing real usage feedback first.

Three questions worth watching

The joint subscription ties three tools together, saving the hassle of paying separately, but several key pieces of information remain to be clarified: the specific quota limits for each tier, the actual numerical value of the shared weekly quota, and the billing method after exceeding the quota. Until these numbers are public, it's hard to assess exactly what the difference is between the $8 entry price and the $200 high-tier price.

For users already using one or two of these products, the core criterion for whether to migrate is not the price itself, but whether the shared quota can cover their real usage combination.

One-line Summary

Grok, Cursor, and X launch a joint subscription from $8 to $200, with Grok invocations sharing a single weekly quota.

Paragraph Summary

Grok, Cursor, and X have been integrated into a single subscription plan, offering a free tier and four paid tiers with monthly fees ranging from $8 to $200 (before tax). The product logic is to string together model capabilities, programming landing scenarios, and social distribution touchpoints into one chain. The most critical detail is that Grok's invocations in the X app and Cursor share the same weekly quota rather than separate allocations, which may cause high-intensity programming scenarios to exhaust the quota faster. Meanwhile, third-party models like Claude and GPT in Cursor are not included in the shared system and require separate payment. Existing subscribers can maintain their original plans, as migration is gradual and non-mandatory. The specific quota limits for each tier and overage billing methods have not yet been announced, so actual cost-effectiveness remains to be verified.