-
Notifications
You must be signed in to change notification settings - Fork 13.5k
Gemini CLI usability issues despite Pro subscription #24712
Copy link
Copy link
Closed as duplicate of#24937
Labels
area/documentationGemini CLI documentation tasks and issuesGemini CLI documentation tasks and issuesarea/extensionsIssues related to Gemini CLI extensions capabilityIssues related to Gemini CLI extensions capabilityarea/platformIssues related to Build infra, Release mgmt, Testing, Eval infra, Capacity, Quota mgmtIssues related to Build infra, Release mgmt, Testing, Eval infra, Capacity, Quota mgmtstatus/need-triageIssues that need to be triaged by the triage automation.Issues that need to be triaged by the triage automation.status/possible-duplicate
Description
Metadata
Metadata
Assignees
Labels
area/documentationGemini CLI documentation tasks and issuesGemini CLI documentation tasks and issuesarea/extensionsIssues related to Gemini CLI extensions capabilityIssues related to Gemini CLI extensions capabilityarea/platformIssues related to Build infra, Release mgmt, Testing, Eval infra, Capacity, Quota mgmtIssues related to Build infra, Release mgmt, Testing, Eval infra, Capacity, Quota mgmtstatus/need-triageIssues that need to be triaged by the triage automation.Issues that need to be triaged by the triage automation.status/possible-duplicate
Type
Fields
Give feedbackNo fields configured for issues without a type.
URL of the page with the issue
https://geminicli.com/
What is the problem?
I am a Pro user and the current experience with Gemini CLI is not reliable for real work. The limits are too easy to hit even during normal usage, which interrupts workflows constantly. On top of that, response times can exceed 5–10 minutes, making it impractical for iterative tasks.
Another major issue is that the system sometimes switches to a different model without clear notice, which leads to inconsistent output quality and behavior. Context handling is also weak — the model frequently loses track of previous instructions or relevant information, forcing me to repeat myself and wasting time.
Overall, the tool feels significantly worse than before and is not meeting expectations for a paid product.
What did you expect to happen?
As a Pro user, I expect:
Stable and predictable limits suitable for professional use
Reasonable response times (not multi-minute delays)
Consistent model behavior without silent switching
Strong context retention across interactions
A reliable tool that improves productivity, not reduces it
Additional context
At this point, the current experience is affecting my workflow and productivity. If these issues are not addressed or terms are not improved, I will consider switching to alternatives like Claude Code or other tools. I am paying for this service, so I expect a level of quality and reliability that justifies the cost.