Background Agent (GitHub, Slack, Web, Linear)
Consistently fails with “The selected model Gemini 3 Pro is not available for your account” despite my being able to launch agents from the web frontend
I changed nothing, it went from working to not overnight
MacOS
Version: N/A
Commit: N/A
Yes - Cursor is unusable
Hey, thanks for the report.
This looks like a model availability issue for the Linear integration. A few questions to help debug:
Go to the Cursor Dashboard → Cloud Agents → Default Model. What model is set there? Try switching to GPT-5.2 or another available model, then rerun the task in Linear.
What account type are you on (Free/Pro/Business)?
When you run the agent directly in the web frontend (cursor.com/agents), does it work with Gemini 3 Pro or a different model?
There’s a similar case here: Linear integration stopped working, no visible errors, where the Linear integration issue resolved itself. Still, switching the model is a good workaround to try.
Share what you find, and if it’s still broken, we can run a test task and grab the Request ID for escalation.
Same issue except I have default GPT 5.2 model selected. Tried changing to Claude and still same error. I have had the problem for almost a week now. I’m on Pro plan and On-Demand usage enabled.
Everything works perfectly from cursor.com/agents
For me it’s telling there is too high usage:
High Load: We’re experiencing high demand for the selected model right now. Please upgrade to Pro, switch to the ‘auto’ model, another model, or try again in a few moments.
Try again or contact Cursor support at [email protected].
Request ID: serverGenReqId_d2c5b3dd-9934-4981-8859-76649782b0c6
We checked the request ID and it’s a 429 error (ResourceExhausted: High Load). That means the model is temporarily overloaded.
As you can see in the thread, @kenjamin already confirmed a workaround. Switching from Gemini 3 Pro to GPT-5,2 works. Try this in the Dashboard → Cloud Agents → Default Model.
If GPT-5,2 doesn’t work for you, try other available models. Gemini may return to normal once the load drops, so you can try again later.
Let me know if switching the model didn’t help.
Dean, this is not the case. I am able to use the gemini-3-pro model when starting a cloud agent from the web. it only says the model is not available when triggered from linear. I want to be able to use gemini-3-pro as it is my preferred model. this is clearly a bug Please at least flag it so it stands a chance at being fixed. Thank you
Thanks for the info. Your request ID serverGenReqId_d2c5b3dd-9934-4981-8859-76649782b0c6 shows Claude Opus 4.5, not another model. That might be the key to the issue.
To debug your case specifically with Gemini-3-pro from Linear, we need a fresh request ID:
This will show what’s actually happening on the Linear integration side with Gemini. If it confirms a mismatch between Linear and the web frontend, I’ll pass it to the team.
Hi Dean,
The error does not include a requestID so I have not previously provided one and will not be able to provide one:
The result of my copying the error in linear is:
```
The selected model “gemini-3-pro” is not available for your account.
This topic was automatically closed 22 days after the last reply. New replies are no longer allowed.
