Grok Bot is xAI's always-on cloud agent product, and the subscription is only the entry ticket. It includes a weekly token allowance of undisclosed size, then bills uncapped overage at raw model rates, with no spend cap and no model picker. Set your access path and workload below. The calculator shows the estimate and the uncertainty xAI leaves unpublished.
The SuperGrok prices are reported figures, not prices officially published by xAI. Cursor prices are list prices checked 2026-08-22. Every route requires a Cursor account, and the Grok-to-Cursor account link is permanent.
Shown only for Cursor Teams plans. Subscription cost is plan price times seats.
One account is capped at 50 Bots, reported by Daily Dose of Data Science and unconfirmed by xAI.
Only input tokens can take the $0.50/M cached-input rate. The rest of input is $2/M; output is $6/M.
Unknowable from public information. xAI confirms a weekly token allowance but does not publish its size. This 20% to 80% slider estimates the share of blended token cost covered, not a plan entitlement.
xAI does not publish the allowance; this is the fixed 20%/80% scenario band.
Monthly usage is bots × active hours per day × 30 days × the intensity preset. Total tokens split 60% input and 40% output.
Base token cost prices uncached input at $2/M, cached input at $0.50/M, and output at $6/M on grok-4.6.
Long prompts are approximated by treating 10% of light, 35% of steady, and 70% of heavy requests as having prompts above 200K tokens. Those requests cost 2x; assuming the same average pre-multiplier cost as shorter requests, the weighted multiplier is 1.10x, 1.35x, or 1.70x.
Overage is raw token cost × the share of blended token cost the allowance does not cover. The fixed 20%/80% scenario band holds everything else constant and calculates the low endpoint at 80% coverage and the high endpoint at 20% coverage.
Continuum costs $0 for the app on hardware you own. Your model cost remains whatever subscriptions you already pay for or whatever per-token provider you choose. That is a different cost shape, not a claim that tokens are free.
Compute stays in xAI's cloud. The tradeoff is an undisclosed allowance, no spend cap, and no model picker.
The workbench runs agents on your own Mac or connected machine and lets you control them from iOS or web. You provide the always-on hardware and pay your existing subscriptions or any provider's per-token rates.
The real tradeoff: Grok Bot includes hosted always-on compute. Continuum uses a machine you own and operate. Hosted compute is simpler when you do not want to keep hardware online; owned compute gives you provider choice and control over where the agent runs.
SuperGrok Plus at about $100/month and SuperGrok Heavy at about $300/month are reported prices, not prices officially published by xAI. They are deliberately marked “reported” everywhere they appear in this calculator.
Primary product references: xAI's Grok Bot overview, xAI's model pricing, and Cursor pricing. Facts were checked 2026-08-22.
| Input | Value used | Status on 2026-08-22 |
|---|---|---|
| SuperGrok access | Plus ~$100/mo · Heavy ~$300/mo | Reported publicly; not officially published by xAI. |
| Cursor individual access | Pro+ $60/mo · Ultra $200/mo | Published plan prices, verified 2026-08-22. |
| Cursor team access | Standard $40 · Premium $120 / seat / mo | Published plan prices, verified 2026-08-22. |
| Weekly token allowance | Included, size undisclosed | xAI does not publish the quantity. User-estimated in this tool. |
| Metered overage | Uncapped · no spend cap · no model picker | Raw model-rate billing after the allowance. |
| Plausible default model | grok-4.6 · $2 in · $0.50 cached in · $6 out / M | Calculator assumption. Rates double when one request's prompt exceeds 200K tokens. |
| Workload presets | 200K · 500K · 1.5M tokens / active hour | Planning assumptions chosen for this tool, not xAI allowances or benchmarks. |
All costs are USD before tax. One month is modelled as 30 active-calendar days. Individual plans are counted once; Cursor Teams plans multiply by the seat count. Bot count scales token usage, not individual-plan subscription count.
Read the full Grok Bot pricing guide or compare the operating models in Grok Bot vs Continuum.
Access starts with an eligible plan: Cursor Teams Standard is $40 per seat per month, Cursor Pro+ is $60, Cursor Teams Premium is $120 per seat, and Cursor Ultra is $200. SuperGrok Plus is reported at about $100 per month and SuperGrok Heavy at about $300, but xAI does not officially publish those two prices. The final monthly cost adds uncapped metered overage after an undisclosed weekly token allowance.
No. After the undisclosed weekly token allowance, Grok Bot uses uncapped metered overage at raw model rates. There is no published spend cap, so an always-on bot can continue adding cost until you stop the work.
Yes, Grok Bot has limited trial usage. The trial duration is not published, so do not budget around any specific trial length.
xAI does not publish its size. That is why this calculator asks what share of blended token cost the allowance might cover and shows a fixed 20 percent/80 percent scenario band.
Grok Bot does not expose a model picker. This calculator uses grok-4.6 as the plausible default, at $2 per million input tokens, $0.50 per million cached input tokens, and $6 per million output tokens, with all three rates doubled for requests whose prompt exceeds 200K tokens.
The selected estimate uses the share of blended token cost covered that you choose, constrained to 20 through 80 percent. The fixed 20 percent/80 percent scenario band holds every other input constant and recomputes those two endpoints. It isolates the largest unknowable input instead of hiding it inside one precise-looking total.
The calculator makes the undisclosed allowance visible as a fixed 20%/80% scenario band. The guides unpack the pricing and compute tradeoffs; Continuum gives you a free workbench for running and remotely controlling agents on hardware you own.
free app · your hardware · your model providers