Repository navigation
Go: deepseek-v4.1-flash meters report $12/$30/$60 while docs and the console table say $20/$50/$100 #61014
Description
Description
A Go subscription's usage meters report lower limits than the documented ones for deepseek-v4.1-flash, by a constant factor of 0.6.
GET /console/api/go/status (with x-org-id) returns:
fiveHour limitMicroCents 1200000000 ($12)
week limitMicroCents 3000000000 ($30)
month limitMicroCents 6000000000 ($60)
The Console renders those as 0% / 30% / 15%, so the percentages are computed from the $60 base.
Docs (opencode.ai/docs/go) and the Console's own model table both say deepseek-v4.1-flash on Go has a monthly limit of $100 — and the docs state 5-hour: 20% / weekly: 50% / monthly: 100%, i.e. $20 / $50 / $100.
So the three reported limits are exactly 0.6× the documented ones ($12/$20, $30/$50, $60/$100): same shape, different base.
Additionally, opencode.ai/go still shows "Previously $15; now $60" (4x Usage badge) for this model.
Steps to reproduce
- Subscribe to Go, then call
GET /console/api/go/statuswith thex-org-idheader. - Compare
access.meters.{fiveHour,week,month}.limitMicroCentsagainst thedeepseek-v4.1-flashmonthly limit onopencode.ai/docs/go. - Ratios come out at 0.6 for all three windows.
Account: a Personal workspace on Go. Details moved to email as requested.
Expected
Either the meters report $20 / $50 / $100, or the docs and console table report $12 / $30 / $60. Right now a user planning against the documented $100 hits the wall at 60% of it.
Related context: PR #60532 noted that a docs change could ship while "the registry and the Console are back at the old limits", so docs/registry drift seems to be a known failure mode here.
Operating System
Windows 11
OpenCode version
Using the Go API from a custom agent (Hermes), not the OpenCode CLI, so billing was checked via the Console UI and the endpoint above.
Activity
opencodejr commentedon Nov 1, 2026
Thanks for the detailed comparison. Seeing your Go meters at exactly 0.6× the documented DeepSeek V4.1 Flash limits is confusing, and you're right that the published numbers don't currently agree with each other.
The limits your Go account actually gets are set on the service side, and the team has to check them for your specific account. Billing and account support isn't handled in GitHub issues, so please email help@anoma.ly with the details (your workspace, the
/go/statusoutput and what you expected). If you've already emailed, the team will get to it, especially now that we've seen it here. Please keep account details, email addresses and payment information out of this public issue and put them in the email instead.Some context that may help:
The bot will close this issue in 1 day.
Code0020161314 commentedon Nov 1, 2026
Correction on my part, and thanks - #60311 changes how I read these numbers.
I reported the meters as if they were a per-model limit. They aren't: it's one plan-level pool where $60 is 100%, split 20% / 50% / 100% into $12 / $30 / $60. That's a plan constant, so comparing it against DeepSeek's documented per-model limit was a category error. The "0.6x" I flagged is just that 20/50/100 split applied to the constant - it is not evidence that my account is provisioned below the documented number. Withdrawing that reading.
One thing I can actually measure, in case it's useful: billed usage for my current period is $4.7355 (Activity -> Usage, window starting at
access.startsAt), while the meters consumed $9.1611 - ratio ~1.93. So usage is counted into the pool at roughly twice billed spend, not 1:1. That's consistent with per-model weighting, but from outside I can't tell what DeepSeek V4.1 Flash's weight is meant to be, or whether shipping the $100 limit (#60629) changes the weight.Taking the account-specific part to help@anoma.ly as you asked, and removing the workspace id from this issue.
opencodejr commentedon Nov 2, 2026
Hi @Code0020161314, I'm closing this now because billing and account questions are handled by email, not in GitHub issues. If the DeepSeek V4.1 Flash limits on your Go account are still unclear, please email help@anoma.ly with the details, and keep account and payment information out of this public thread. If you've already written to them, the team will get to it. For updates on the published limits, see #60629 and #60311. Thanks again for the thorough write-up.