# Is Grok Bot down? How to check, and what to do

Grok Bot's status is the Grok Bot row on status.cursor.com. See what an outage looks like in the app, what to do while it lasts, and how to tell it's only you.

*From Learn Grok Bot, an independent guide (not affiliated with SpaceXAI or Cursor). Web version: https://learngrokbot.com/is-grok-bot-down Last verified: 2026-10-06.*

To see whether Grok Bot is down, open [status.cursor.com](https://status.cursor.com) and read the row named **Grok Bot**. The app doesn't show service status, so that page is the place to check. If an incident is open, wait until it's marked **Resolved**, and leave the app and the computer alone until then.

**In short**

- Grok Bot's status is on Cursor's status page, in the **Grok Bot** row. SpaceXAI's status page covers the Grok app and has no row for Grok Bot.
- Inside the app an incident looks like ordinary errors: **Bot failed to respond**, messages that won't send, a computer that can't be reached, routines that don't run.
- While an incident is open, wait. Don't reset the computer and don't reinstall the app.
- Afterwards, send your message again, and run by hand any routine that missed its time.
- From 2026-08-18 to 2026-10-06 the page listed 21 incidents that named Grok Bot. Twelve of the 20 that ended were over within two hours.

## Check the Grok Bot row on the status page

Cursor runs the service behind Grok Bot, so Grok Bot's status is on Cursor's status page. Open [status.cursor.com](https://status.cursor.com) in a browser and scroll to the row named **Grok Bot**.

![The status.cursor.com page in a phone browser. A row named Grok Bot reads Operational in green, over a strip of daily bars, most green and a few red or yellow. Under the heading Past Incidents, the entry for Sep 16, 2026 is titled Investigating service degradation, Grok Bot, with three updates: Resolved at 20:57 UTC, Identified at 19:51 UTC and Investigating at 14:27 UTC](https://learngrokbot.com/assets/shots/is-grok-bot-down-status-page.png)

1. The row's state. **Operational** means no incident is open. The other states are **Degraded Performance**, **Partial Outage**, **Major Outage** and **Maintenance**.
2. One bar for each day. A red or yellow bar is a day that had an outage.
3. Incidents are listed under the rows, each with its updates, newest first. **Resolved** is the update that ends one.

Image: Cursor forum (https://forum.cursor.com/t/t-f86577-cannot-connect-on-windows-macos-and-iphone-since-11-sep/171962/1)

- **Read the row, and open any incident in progress.** An incident's title doesn't always name Grok Bot. Seven of the 21 so far were titled for a model, for another Cursor product or for nothing in particular, and listed Grok Bot among the parts affected.
- **Subscribe if you rely on your Bots.** The page has a subscribe button, which sends you its updates.
- **Skip status.x.ai.** SpaceXAI's own status page covers Grok the chat app, the API and x.ai's sites. It has no row for Grok Bot, so a green page there says nothing about your Bots.

## What an incident looks like inside the app

During an incident the app shows the same messages it shows for any failure, and the one most people meet is this banner.

![The left half of a dark red banner with a warning triangle. Its title is Bot failed to respond, and the line under it reads Something went wrong while generating the reply. Try again.](https://learngrokbot.com/assets/shots/is-grok-bot-down-banner.png)

*The banner reads the same whether the cause is an incident or something on your side, which is why the status page comes first.*

Image: @AmericanIrish on X (https://x.com/AmericanIrish/status/2107477178053271867)

Here are four of the status page's own wordings, each with the date it appeared, beside what that looks like in the app.

| On the status page | In the app |
|---|---|
| “Users may experience degraded performance or bots that don't respond.” (2026-09-16) | **Bot failed to respond** on every Bot, slow replies, or messages marked **Failed to send** |
| “Some users may be unable to reach their Grok Bot computer, and Recover or Reset may fail.” (2026-09-07) | **Reconnecting**, then “Can’t reach your computer”. **Recover failed** or **Reset failed** if you try either one |
| “Some trigger fires may be dropped.” (2026-09-03) | A routine doesn't run when its time or its trigger comes |
| “an issue causing Grok Bot connectors to hang” (2026-08-18) | A Bot stalls on a step that uses a plugin |

Each of those messages has its own entry in [Error messages](https://learngrokbot.com/error-messages).

## What to do while an incident is open

An incident is a fault in the service that sits between your app and your Bots' computer. Nothing you do at either end reaches it.

Diagram: Three parts in a row: the app on your device, Grok Bot's service in the middle, and the cloud computer your Bots work on. The service is hatched: during an incident the fault is there. A note over the app reads reinstalling won't help, and a note under the computer reads a reset can lose recent work.
- Your app → Grok Bot's service
- Grok Bot's service → Your Bots' computer
- Note on Your app: reinstalling won't help
- Note on Your Bots' computer: a reset can lose recent work

*The hatched part is where an incident happens. The two notes are what people try at each end.*

- **Wait for Resolved.** Keep the status page open, or subscribe to it. The incident is over when the page marks it **Resolved**.
- **Don't reset the computer.** A reset doesn't fix a service fault, and it can lose very recent changes when the computer can't be reached to save them first.
- **Don't reinstall the app.** Signing out and making a new Bot don't help either.
- **Send your message again afterwards.** A message sent during an incident isn't retried for you.
- **Run missed routines by hand.** A routine due during the incident of 2026-10-03 didn't run, and on 2026-09-03 some trigger fires were dropped. Check your routines, and ask a Bot to run any that missed its time.

Your chats are stored apart from the computer, so they're there when the service comes back.

## How long incidents have lasted

From 2026-08-18 to 2026-10-06 the status page listed 21 incidents that named Grok Bot. Twenty of them have an end time, and the chart shows how long each one ran.

Diagram: A bar chart of 20 Grok Bot incidents in date order, from Aug 18 to Oct 3, 2026. Bar height is how long each lasted, on a scale marked at 2, 4 and 6 hours. Most bars are under 2 hours. The tallest, on Sep 16, is labelled 6 h 30 min, and the shortest, on Sep 28, is labelled 13 min. Pale bars were rated minor and solid bars were rated major or critical.

2 h
4 h
6 h
Aug 18
Sep 3
Sep 16
Sep 25
Oct 3
6 h 30 min
13 min
Rated minor
Rated major or critical

*Each bar is one incident, oldest on the left. The times are the status page's own, from its first update to **Resolved**, in UTC.*

- Seven lasted an hour or less. The shortest, on 2026-09-28, took 13 minutes.
- Twelve were over within two hours.
- Three ran longer than four hours, all between 2026-09-16 and 2026-09-18. The longest, on 2026-09-16, took 6 hours 30 minutes.
- The page rated eight of the 21 major and one critical. The 21st opened on 2026-10-06 at 19:27 UTC.

How long earlier incidents ran doesn't set the length of the next one. The updates on the status page are the only measure of one that's open.

## Tell an outage from a problem on your side

When the row reads **Operational** and a Bot still won't answer, three checks sort it out. Take them in order, and stop at the first one that gives you an answer.

### Step 1: Look at the Grok Bot row again

An incident in progress means the problem isn't yours: [wait it out](#wait). The page can trail the start of a problem. On 2026-10-03 the fault began at about 06:00 UTC, and the incident was posted at 06:50 UTC. So when every Bot failed at once a few minutes ago, look again shortly before you change anything.

### Step 2: Message a different Bot, in a new chat

Send it one word. If that Bot answers, the service is up, and the problem is in one chat or one Bot: go to [Bot not responding](https://learngrokbot.com/not-responding#resend). If every Bot fails, go on to the third check.

### Step 3: Open Grok Bot on your phone, over mobile data

Turn off Wi-Fi on the phone and sign in with the same Cursor account. If your Bots answer there, the service and the computer are both fine, and your usual network is in the way: see [the network fixes](https://learngrokbot.com/cant-reach-computer#network).

Still failing everywhere, with the row at **Operational**? Then the cause is on your account: your weekly usage has run out, or the computer your Bots share is stuck. [Bot not responding](https://learngrokbot.com/not-responding#usage) takes those two in order.

**Sort it: Outage, or on your side? Sort each situation.**

Answer:
- The Grok Bot row reads “Partial Outage” → Wait and watch the status page (Any state other than Operational means an incident is in progress.)
- An incident titled for xAI models lists Grok Bot as affected → Wait and watch the status page (The title doesn't have to name Grok Bot. The list of affected parts does.)
- Every Bot failed at once two minutes ago, and the row still reads “Operational” → Wait and watch the status page (The page can trail a problem. Look again shortly before you change anything.)
- One Bot is silent, and your other Bots answer → Fix it on your side (The service is up. Look in that Bot's chat for a card it's waiting on.)
- Your Bots answer on your phone over mobile data, but not on your laptop → Fix it on your side (The service and the computer are fine. Your network is blocking the app.)
- Weekly usage reads 100% → Fix it on your side (Your Bots have stopped until the reset. An incident has nothing to do with it.)

With an outage ruled out, the next page has the fixes for a Bot that still won't answer.

**Next:** [Bot not responding](https://learngrokbot.com/not-responding)

No incident, and still no reply? Work through the fixes in order, from sending the message again to the one reset worth doing.

### Cursor is having an incident. Is Grok Bot affected?

Only if the **Grok Bot** row says so. Each of Cursor's services has its own row on the status page, so the Cursor editor can have an incident while Grok Bot's row stays **Operational**. One other row matters to some Bots: **Cloud Agents**. A Bot that hands coding work to a Cursor Cloud Agent depends on that service too.

### Where do I report a problem when no incident is listed?

Email hi@cursor.com with the exact message, your version and the request ID. [Bot not responding](https://learngrokbot.com/not-responding#support) shows where each of those comes from.

- 2026-10-06: First published.

## Sources

- [Cursor Help Center: Grok Bot FAQs (Is Grok Bot down?)](https://cursor.com/help/grok-bot/faqs)
- [Cursor status page (the Grok Bot row and its states)](https://status.cursor.com/)
- [Cursor status page: incident history (the 21 incidents that name Grok Bot, 2026-08-18 to 2026-10-06)](https://status.cursor.com/history)
- [SpaceXAI status page (Grok, API and Services; no Grok Bot row)](https://status.x.ai/)
- [Cursor Help Center: Grok Bot computer (what a reset can lose, where chats are stored)](https://cursor.com/help/grok-bot/computer-recovery)
- [SpaceXAI: Grok Bot changelog (Failed to send)](https://x.ai/changelog/bot)
- [Cursor forum: staff on resending messages and running missed routines after the incident of 2026-10-03](https://forum.cursor.com/t/grok-bots-failing-to-respond-again/173658)
- [Cursor forum: staff on messages that failed to send from 2026-09-16](https://forum.cursor.com/t/some-of-the-bots-became-unresponsive-failed-to-send-on-my-messages-to-them/171858)
- [Cursor forum: staff on why quitting and signing out don't help during a service fault](https://forum.cursor.com/t/model-name-is-not-valid-grok-4-7-0907-grok-bot/170932)
- [Image: the status page with the Grok Bot row and an incident (Cursor forum)](https://forum.cursor.com/t/t-f86577-cannot-connect-on-windows-macos-and-iphone-since-11-sep/171962/1)
- [Image: the Bot failed to respond banner (@AmericanIrish on X)](https://x.com/AmericanIrish/status/2107477178053271867)