You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
OpenCode Go documents usage limits per model, but once the workspace weekly
meter reaches 100%, every Go model is blocked, including models that were
never used. Switching models does not work around the limit.
Plugins
No response
OpenCode version
2.2.6
Steps to reproduce
Subscribe to OpenCode Go (Personal workspace, single member).
Use one Go model (opencode-go/deepseek-v4-pro) until GET /api/experimental/usage/limits reports go.weekly.percent = 100 and go.weekly.limited = true.
At that point go.monthly.percent is 50.
Switch the session with /models to a different Go model that has never
been used (opencode-go/minimax-m3).
Thanks for the clear report. I think this is a duplicate of #60311.
OpenCode Go has one allowance per workspace, split into 5-hour, weekly and monthly windows and shared by all paid models. Each model's listed monthly limit only sets how heavily its usage counts against that shared pool. Once any window reaches 100%, every paid Go model is blocked, including ones you haven't used; free models keep working. So switching from deepseek-v4-pro to minimax-m3 can't get around the weekly limit. You're right that the docs still describe the limits as per model, which is misleading, and #60311 tracks exactly that mismatch.
Possible duplicates and related issues, ordered from most to least likely:
The bot will close this issue as a duplicate in 1 day unless you reply explaining how your problem is different. If you'd like to follow progress, please subscribe to #60311.
We haven't heard back, so I'm closing this as a duplicate of #60311. That issue reports the same thing: once any Go window (5-hour, weekly or monthly) reaches 100%, every paid Go model is blocked, including ones you haven't used. Go has one shared allowance for all paid models, but the docs describe the limits as per model. Open PR #61406 rewrites the docs to describe the shared allowance.
Please subscribe to #60311 to follow it. If your problem turns out to be different, reply here and explain how, and we can reopen this.
Activity
opencodejr commentedon Nov 7, 2026
Thanks for the clear report. I think this is a duplicate of #60311.
OpenCode Go has one allowance per workspace, split into 5-hour, weekly and monthly windows and shared by all paid models. Each model's listed monthly limit only sets how heavily its usage counts against that shared pool. Once any window reaches 100%, every paid Go model is blocked, including ones you haven't used; free models keep working. So switching from
deepseek-v4-protominimax-m3can't get around the weekly limit. You're right that the docs still describe the limits as per model, which is misleading, and #60311 tracks exactly that mismatch.Possible duplicates and related issues, ordered from most to least likely:
The bot will close this issue as a duplicate in 1 day unless you reply explaining how your problem is different. If you'd like to follow progress, please subscribe to #60311.
opencodejr commentedon Nov 8, 2026
We haven't heard back, so I'm closing this as a duplicate of #60311. That issue reports the same thing: once any Go window (5-hour, weekly or monthly) reaches 100%, every paid Go model is blocked, including ones you haven't used. Go has one shared allowance for all paid models, but the docs describe the limits as per model. Open PR #61406 rewrites the docs to describe the shared allowance.
Please subscribe to #60311 to follow it. If your problem turns out to be different, reply here and explain how, and we can reopen this.