● SuperGrok is retired Replaced by The X Subscription on 12 October: Free, Plus $8, Premium $30, Super $100, Ultra $200.
● One pool, three products Paid plans share one usage pool across the Grok apps, the Grok tab on X, and Cursor.
● Current models Grok 4.7 at $2/$6 per million tokens, and Grok Voice 3.0 at $0.08 per minute.
Updated through November. Last checked 1 November 2026.
If you are looking for SuperGrok, SuperGrok Plus or SuperGrok Heavy, those tier names no longer describe what xAI sells. This page tracks what is actually live, with every figure read from x.ai rather than from coverage of it.
The X Subscription: what replaced SuperGrok
On 12 October 2026 xAI introduced The X Subscription, a single plan bundling Grok, Grok Bot, Cursor and X Premium. Existing SuperGrok, Cursor and X Premium subscribers could keep their plan or switch, with a prorated refund on the old one.
| Tier | Price | Usage | What is included |
|---|
| Free | $0 | Limited | Daily Grok access, a starter Cursor coding-agent allowance, standard X |
| Plus | $8/mo | Base | More Grok and Cursor usage, X checkmark, creator tools, fewer ads |
| Premium | $30/mo | 4x Plus | SuperGrok plus Cursor Pro plus X Premium+ benefits - xAI puts the combined former price at $90/mo |
| Super | $100/mo | 15x Plus | All of the above, ads removed from the X timeline |
| Ultra | $200/mo | 50x Plus | All of the above |
The pooling is the part that changes how you budget. Paid plans share one pool of Grok model usage across the Grok apps, the Grok tab on X, and Cursor. Each plan includes at least twice as much coding usage with Grok models as the comparable old Cursor plan. Third-party models in Cursor are available on Premium and above, and anything beyond the pool bills separately at API list price through top-ups.
Team plans carry the same per-seat prices as the individual Premium, Super and Ultra tiers, with three times the coding usage of the old Cursor Teams plans, plus SSO, admin controls, consolidated billing, usage analytics and SOC 2 compliance. Enterprise manages Grok and Cursor from one dashboard on a single usage pool.
Worth checking before you switch: the x.ai pricing page comparison table still carries the legacy column headings SuperGrok Lite, SuperGrok and SuperGrok Heavy, and prices only SuperGrok at $30. The announcement and the comparison table have not fully converged, so confirm what your account is actually on rather than assuming from either page alone.
Grok 4.7: the current flagship
Grok 4.7 shipped on 21 September 2026, built on a larger base model than Grok 4.6 with a longer reinforcement learning run. xAI describes it as its most capable model for coding and knowledge work.
| Benchmark | Grok 4.7 xHigh | Grok 4.6 High |
|---|
| Terminal-Bench 4.0 | 37.6% | 20.3% |
| CursorBench 4.0 | 46.3% | 40.4% |
| DeepSWE v1.1 | 71.0% (high effort) | 65.2% |
| EEBench | 64.0% | 53.0% |
| HealthBench Professional | 56.7% | 48.5% |
| Harvey Legal Agent Benchmark | 19.6% | 15.8% |
| GDPval (Elo) | 1,605 | 1,542 |
| AA Briefcase v1.1 | 1,657 | 1,546 |
The Terminal-Bench jump is the one to notice: 20.3% to 37.6% is close to a doubling on agentic terminal work, where the other gains are incremental. These are xAI figures and have not been independently validated.
API pricing: $2 per million input tokens and $6 per million output. A fast variant runs at twice the output speed for twice the price. On safety, xAI reports Grok 4.7 allows 3.3% of risky dual-use prompts through on HackerBench v0.3, and scores 62.4% on the LatchBio biosafety benchmark.
Grok 4.7 is available in Cursor, Grok Build, the Grok API, third-party coding harnesses and model routers. It also reached GitHub Copilot on 21 September, on Copilot Pro, Pro+, Max, Business and Enterprise, billed at provider list price through usage-based billing. Business and Enterprise admins can disable it through the Copilot model policy - under default enablement new models turn on automatically.
Grok Voice 3.0: the cheapest credible speech-to-speech model
Grok Voice 3.0 landed on 14 October 2026 at $0.08 per minute, or $4.80 an hour - the same price as Voice 2.0. xAI reports it first on the Artificial Analysis Speech-to-Speech Index at 86.9, up from 81.3.
- Full Duplex Bench, which measures pauses, turn-taking and interruptions: 97%
- Big Bench Audio, speech reasoning: 98.5%
- Task success rate: 94.8%
- EVA-Bench: 74%, which xAI reports as the highest recorded
- Latency under one second, with τ-Voice latency 40% below GPT-Live-1 (Astra)
The cost comparison xAI publishes is more useful than the benchmark table: average cost per task on Big Bench Audio is $0.025, against $0.030 for GPT-Live-1 (Astra, Medium) and $0.055 for GPT-Realtime-2.1 (High). That is roughly half the cost of the most expensive comparison point for a task of the same shape.
It powers voice mode in Grok Bot and handles conversations in Tesla vehicles. xAI also says it runs Starlink call centres handling 20,000 calls a day across multiple languages, with about 70% resolved without a human.
API changes you may need to action
| Date | Change | What to do |
|---|
| 27 Oct | Custom moderation controls | Add a guidelines array to Chat Completions or Responses requests to set handling for the nsfw, violence and profanity categories |
| 14 Oct | Grok Voice 3.0 available | grok-voice-latest now points to it. Agents on that alias were switched automatically and need no change |
| 14 Oct | Grok Voice 2.0 deprecated | Set model to grok-voice-3.0 or grok-voice-latest. Same price |
| 2 Oct | grok-voice-transcribe-1.0 end of life | Requests now route to grok-voice-transcribe-2.0 at the same price |
If you pin model IDs rather than using the latest aliases, the two deprecations above are the ones that will bite silently.
Grok Bot runs on its own meter
Grok Bot, a set of always-on agents that work inside tools and apps, launched in beta on 11 August and expanded to all SuperGrok and Cursor plans on 26 August. The detail that matters for cost planning: Grok Bot has its own usage pool, separate from your Grok and Cursor plans, so work handed to a Bot does not draw down the shared pool.
What has shipped in November so far
Nothing yet. As of 1 November 2026, xAI has published no November items and the API release notes were last updated on 27 October. This section is updated as releases land, so the page reflects the month rather than predicting it.
Which tier, if any
- You only use Grok in the app → Plus at $8 is the cheapest paid entry and the biggest relative jump from Free.
- You code in Cursor and use Grok → Premium at $30. xAI puts the combined former price of those products at $90, and it is one pool rather than two bills.
- You were on SuperGrok Heavy → compare against Super at $100 and Ultra at $200 on usage multiplier, not on tier name. The names do not map across.
- You build on the API → the subscription is irrelevant. Grok 4.7 is $2/$6 per million and Grok API usage is billed separately from the plan pool.
- You run voice agents → Grok Voice 3.0 at $0.08/min is the cheapest of the three models xAI benchmarks against on cost per task.
FAQ
Is SuperGrok discontinued?
The tier names were replaced by The X Subscription on 12 October 2026. Existing SuperGrok subscribers could keep their plan or switch with a prorated refund. The x.ai pricing comparison table still shows SuperGrok headings, so check your own account rather than the marketing page.
How much does Grok cost per month in November 2026?
Free is $0, Plus is $8, Premium is $30, Super is $100 and Ultra is $200. API use is billed separately at $2 per million input tokens and $6 per million output for Grok 4.7.
Does Cursor usage come out of my Grok allowance?
Yes. Paid plans share one pool of Grok model usage across the Grok apps, the Grok tab on X, and Cursor. Grok Bot is the exception and runs on a separate pool.
What is the newest Grok model?
Grok 4.7, released 21 September 2026, for text and coding. Grok Voice 3.0, released 14 October 2026, for speech to speech.
Is Grok 4.7 available in GitHub Copilot?
Yes, since 21 September 2026, on Copilot Pro, Pro+, Max, Business and Enterprise, billed through usage-based billing at provider list price.
Do I need to change anything in my API code?
If you pin grok-voice-2.0 or grok-voice-transcribe-1.0, yes - both are retired. Both replacements cost the same.
Related reading
Sources
Every figure on this page was read from x.ai, docs.x.ai or the GitHub changelog on 1 November 2026. Benchmark scores are xAI figures unless stated otherwise and have not been independently validated. Pricing changes often - check the linked page before you buy.