This isn’t a bug — the usage limit is being triggered because of the model you’re using, combined with how billing works on your plan.
Your team is on the Teams plan with $20/member/month of included API usage. Usage is shared across all your tabs and windows — it’s tracked per user, not per tab. So opening multiple tabs doesn’t create separate usage pools.
The key factor here is model selection. You’re using claude-4.6-opus-high-thinking in Max Mode, which is one of the most expensive models available. A single Max Mode request with this model can consume several dollars of your usage budget due to the high per-token cost (especially for thinking/output tokens). At that rate, $20 of included usage can be exhausted in roughly 10-20 requests, which matches what you’re seeing.
A few options to get more out of your usage budget:
Use Auto mode for everyday tasks — it selects cost-effective models and draws from a separate, more generous usage pool
Use Max Mode selectively — reserve it for complex tasks that genuinely need the larger context window, as the Max Mode docs recommend
Enable on-demand usage — your team admin can set a spending limit to allow pay-as-you-go usage beyond the included $20
Try less expensive models — models like Gemini 3 Flash or Composer 2 cost significantly less per request
