On October 12, 2026, xAI announced The X Subscription — a single subscription that combines the Grok chat, Grok Bot, the Cursor code editor, and X Premium. Plans start at $8 per month, and paid plans feature a shared usage pool for Grok models, available across the chat, social network, and code editor. Subscriptions are no longer sold individually to new customers.
What happened
The announcement was made on October 12, 2026. The lineup includes four plans: Plus at $8 per month, Premium at $30 (4x Plus), Super at $100 (15x Plus), and Ultra at $200 (50x Plus); usage limits vary by up to fifty times between them. Premium replaces the previous SuperGrok + Cursor Pro + X Premium+ bundle, which totaled about $90 per month. Paid plans introduce a shared usage pool for Grok models with a weekly quota: unused volume can be dynamically reallocated between coding and chat, and when credits are exhausted, they can be purchased directly. The pool works in the Grok app, the Grok tab on X, and the Cursor editor. Third-party models Claude and GPT are available in Cursor starting with the Premium plan and are billed separately at API pricing via top-ups. Existing SuperGrok, Cursor, and X Premium subscribers were offered the option to keep their current plan or switch to The X Subscription with a proportional refund.
Context
There is no technical novelty in the announcement: no new model, architecture, training method, or benchmarks — it is a product packaging of an existing stack. The point of the move is different: xAI has for the first time folded the chat, IDE, and social network into a single bundle with a portable pool of compute. Architecturally, this looks like a single consumption counter on top of a shared serving layer — a new UX pattern of a “single token wallet,” where one profile, one billing, and one quota work across heterogeneous products. Such an announcement should be read as a signal: access to the model is becoming a commodity layer, and competition is shifting from the models themselves to packaging, billing, and ecosystems.
Why this matters for the industry
For the industry, the event is significant for its commercial model, not its technology. xAI is consolidating the consumer AI stack — chat, coding agents, and a social network — into a single subscription with a shared token pool: this is a direct price hit on Anthropic’s and OpenAI’s separate subscriptions and a bet on the “one subscription for everything” format, in which the company is willing to trade margin for user retention and cross-product consumption. For builders, the announcement provides no direct lever: it contains no new API, no developer access, and no infrastructure product, so it gives third-party developers nothing in itself. What is worth dissecting is the metering and monetization scheme: one billing and one quota instead of separate limits for each product — a template for your own products. Everything else depends on retention: if the shared pool shows high cross-product consumption, xAI should be expected to reveal detailed limits and fair use rules, and competitors to respond with their own bundle offers; then engineers will have to compare not individual tools, but entire ecosystems.
Why this matters for users
Users of the Grok and Cursor combo have a direct reason to recalculate their budget: the $30 per month Premium plan covers what previously required three subscriptions — SuperGrok, Cursor Pro, and X Premium+ — totaling about $90. The savings only work on xAI models: Claude and GPT in Cursor are still billed separately at API prices, so no savings on third-party models should be expected. It is stated that each plan provides roughly twice as much coding usage of xAI models as a comparable standalone Cursor plan, but the pool is shared across apps, so part of the quota may go where it is not needed. The practical action today is to compare your own usage profile — that is, the volume of coding in Cursor and the number of requests to Grok — with the plan limits at x.ai/pricing and calculate the break-even point. Existing subscribers have nothing to fear: current plans can be kept, and a proportional refund is due upon switching; if desired, accounts can be merged into a single work profile. For those who use only one product in the bundle, migration may not be worthwhile.
What is still unknown / limitations
Absolute pool limits by plan, model versions, and fair use rules were not disclosed in the announcement, so plans cannot yet be compared in detail based on independent data. The claim of twice the coding usage compared to a comparable Cursor plan is given without methodology: the “comparable plan” is not defined, units of measurement — tokens, requests, or windows — are not specified, and measurement conditions are not stated, so this is a marketing comparison, not a reproducible fact. The reason for the threefold price reduction of the bundle — real inference economics or a deliberate investment in retention — is not established by available sources. Finally, it is unknown whether xAI will reveal detailed limits under increased load and whether competitors will respond with their own bundles.
Sources
Author
Look at AI, editorial team