Request consumption appears abnormally high — possible double/triple counting
- Dominant language
- Shell
- Stars
- 11.2k
- Forks
- 1.9k
- Avg merge
- 14h 16m
- Merged PRs (30d)
- 6
Description
### Describe the bug
I'm using the 1x multiplier model (GPT-5.4), but request consumption is behaving
as if a 3x multiplier model is selected. Each interaction burns through quota at
roughly 3x the expected rate, despite explicitly choosing the lower-cost model tier.
This started approximately 3 days ago with no changes on my end, same model
selection, same workflows, same shell environment.
### Affected version
GitHub Copilot CLI 1.0.22.
### Steps to reproduce the behavior
Set model to GPT-5.4
Run a typical planing - implementing workflow, says costs 1x req, but feels like 3x, %'s are increasing way too fast.
### Expected behavior
Expected: 1 request deducted per interaction (1x model)
Actual: ~3 requests deducted per interaction — consumption matches a 3x model
### Additional context
_No response_
Contributor guide
Assessment
This issue has not been assessed yet.