OpenCode Go weekly/monthly quota does not match usage history

Description

Description

My OpenCode Go quota usage does not appear to match the usage shown in Usage History.

I only started using OpenCode Go on August 7, 2026, so there should be no earlier Go usage within my current weekly/monthly period.

My Usage History currently shows:

August 7

DeepSeek V4 Pro (Go): ~$2.01
GPT-5.6 Luna (Go): ~$6.46

August 8

DeepSeek V4 Pro (Go): ~$0.04
Qwen3.8 Max (Go): ~$2.58

Total visible Go usage: approximately $11.09

However, my quota dashboard shows:

Rolling Usage: 87%
Weekly Usage: 100%
Monthly Usage: 52%

If the published Go limits are $30 weekly and $60 monthly, $11.09 of usage would correspond to approximately:

Weekly: 37%
Monthly: 18.5%

This does not appear to match the quota dashboard.

I also noticed unusual accounting on Qwen3.8 Max within the same session:

474,625 input / 6,072 output → $1.2228
480,782 input / 2,237 output → $0.1481
482,809 input / 169 output → $1.2079

These requests were made consecutively in the same session with very similar input context sizes, but the reported costs differ significantly.

Could you please check:

1. Why does approximately $11.09 of visible usage result in 100% weekly and 52% monthly quota usage?
2. Are the dollar amounts shown in Usage History directly used to calculate Go quota?
3. Are there model-specific multipliers or usage that are not shown in Usage History?
4. Was prompt caching applied differently to the three Qwen3.8 Max requests above?
5. Could there be an issue with Go quota accounting for my account?

I can provide additional session/usage information privately if required.

Plugins

No response

OpenCode version

No response

Steps to reproduce

No response

Screenshot and/or share link

Operating System

No response

Terminal

No response

Start by comparing the OpenCode Go quota dashboard with the Usage History entries and the three consecutive Qwen3.8 Max requests described in the issue. No repository files, tests, version, or reproducible steps are identified; done means explaining the discrepancy and confirming whether caching, model multipliers, or unshown usage account for the quota values.