Elevated errors hitting Grok models in Cursor tonight

If requests to Grok models in Cursor fail with a capacity error tonight, Cursor says "Retrying the request or switching to another model may help." Its status page says "Some requests may fail with a capacity error" and, since 6:58 PM ET, "A fix has been implemented and we are monitoring for service restoration." As of 8:30 PM ET the same page lists Grok Bot as Operational.

Update, Oct 16, 1:20 AM ET: Cursor marked the incident resolved: "This incident has been resolved. Grok models in Cursor are working normally again." The status page was All Systems Operational at 7 AM ET. Cursor did not name which Grok models were hit.

• Status: Resolved at 1:20 AM ET on Oct 16 (05:20 UTC). Impact: minor.
• Window start: Investigating posted at 6:25 PM ET (22:25 UTC). Monitoring update at 6:58 PM ET (22:58 UTC).
• Symptoms: Elevated errors on Grok models; capacity errors on some requests.
• Affected (degraded): CLI, Cloud Agents, IDE.
• Not affected (Operational): Automations, Review Agents, cursor.com, Origin, Grok Bot.
• Workaround (Cursor): "Retrying the request or switching to another model may help."

What changed#

At 6:25 PM ET Cursor opened incident h76w09vs1vqq titled "Elevated errors affecting Grok models," with an Investigating note that some requests may fail with a capacity error. At 6:58 PM ET it moved to Monitoring: "A fix has been implemented and we are monitoring for service restoration." Our 6:14 PM ET breaking scan still showed All Systems Operational, so this window opened after that check.

The incident names Grok models in Cursor and marks CLI, Cloud Agents and IDE degraded. It does not list Grok Bot, which the components list shows as Operational.

Why it matters#

Our read: if you run Grok models from the IDE, CLI or Cloud Agents, capacity errors tonight are likely this incident rather than your plan or key. The page does not list Grok Bot as affected, so treat a Bot failure as a separate problem unless the page later marks Grok Bot degraded.

How to use it#

1. Open status.cursor.com/incidents/h76w09vs1vqq and confirm the incident is still Monitoring or resolved.
2. If a request on a Grok model fails, retry it or switch to another model, as the incident suggests.
3. Our suggestion: for Cloud Agents, change the model on the next launch.

Prompt (ours; audit only, it will not clear capacity errors):

Check status.cursor.com for incident h76w09vs1vqq. Summarize whether CLI, Cloud Agents and IDE are still degraded, and whether Grok Bot is still Operational.

Good to know#

• Shortlink for the same incident: https://stspg.io/nbsrtgp6pr7p.
• Impact is listed as minor. The overview banner reads Partially Degraded Service, not a full outage.
• Automations, Review Agents, cursor.com and Origin are listed Operational alongside Grok Bot.
• The timeline and component list come from the status API, rechecked at 8:30 PM ET.

What to watch#

Whether the incident flips to Resolved, whether Grok Bot stays Operational the whole window, and whether Cursor names which Grok model versions were hit.