ERROR_PROVIDER_ERROR for 2 days

Request ID: 27405125-a44b-4e7f-9fbd-3c9f148977b3

Happening on every agent request for 2 days straight. Says “isRetryable”: false
and “might be temporary” but it’s been 2 days, not temporary.

ERROR_PROVIDER_ERROR
{"error":"ERROR_PROVIDER_ERROR","details":{"title":"Provider Error","detail":"We're
having trouble connecting to the model provider. This might be temporary - please
try again in a moment.","isRetryable":false,"additionalInfo":{},"buttons":
,
"planChoices":
},"isExpected":true}

NonRetriableError: Provider Error We're having trouble connecting to the model provider.
at async fH.runAgentLoop (workbench.desktop.main.js:21249:3041)
at async jXm.streamFromAgentBackend (workbench.desktop.main.js:24426:13900)
at async jXm.getAgentStreamResponse (workbench.desktop.main.js:24426:22743)
at async DIe.submitChatMaybeAbortCurrent (workbench.desktop.main.js:12054:1967)

Version: 3.8.11
OS: Mac OS

Version: 3.8.11
VS Code Extension API: 1.105.1
Commit: e56ad3440df06d22ca7501e65fd518e905486ef0
Date: 2026-06-18T01:40:18.333Z
Layout: editor
Build Type: Stable
Release Track: Default
Electron: 40.10.3
Chromium: 144.0.7559.236
Node.js: 24.15.0
V8: 14.4.258.32-electron.0
xterm.js: 6.1.0-beta.256
OS: Darwin x64 23.6.0

Cursor just got bought by SpaceX for $60 BILLION dollars, and I have been completely blocked for two days by this:

ERROR_PROVIDER_ERROR {“error”:“ERROR_PROVIDER_ERROR”,“details”:{“title”:“Provider Error”,“detail”:“We’re having trouble connecting to the model provider. This might be temporary - please try again in a moment.”,“isRetryable”:false,“additionalInfo”:{},“buttons”:[], “planChoices”:[]},“isExpected”:true}

Request ID: 27405125-a44b-4e7f-9fbd-3c9f148977b3
Mac OS
Cursor Version: 3.8.11

“isRetryable”: false. “This might be temporary.” It is not temporary. It has been
TWO DAYS. Every single agent request just dies with this same error, no matter
what I do.

And here’s the part that actually makes me furious: there is nowhere to escalate
this. No support ticket system. No live chat. No status page acknowledging an
outage. Nothing. A company valued at sixty billion dollars, with a customer base
that includes a huge chunk of the Fortune 500, and the entire support channel
for a paying customer whose work is dead in the water is “post on the forum and
hope someone notices.”

This isn’t a $5/month side project. People are running their actual jobs through
this tool. When it breaks, work stops, full stop. At this valuation there is
no excuse for “we don’t really have support.” That’s not a scrappy-startup
quirk anymore, that’s just neglect.

If anyone has a workaround, I’d genuinely appreciate it, because right now my
only support channel is shouting into this thread.

Hi there,

While many users get support through the forum, you’re also welcome to email [email protected] or fill out the contact form at Cursor Help Center.

Based on the Request ID, it looks like you tried sending an unsupported image type to GPT 5.3 Codex. The supported formats are: image/jpeg, image/png, image/gif, and image/webp.

Does that match what you are seeing? I agree the error message should be clearer, and I’ll pass that feedback along to the team.

In the meantime, you can work around this by either saving the image as a file and @-referencing it, or converting it to PNG or JPEG before attaching. Either approach should resolve the error for now.

Let me know if you run into any other issues!