Problem or use case
Orca tracks usage for Claude, Codex, Gemini, OpenCode Go, Kimi, MiniMax, Antigravity and Grok, but not for Z.ai's GLM Coding Plan. Plenty of people run GLM through Claude Code (ANTHROPIC_BASE_URL=https://api.z.ai/api/anthropic) or OpenCode; that traffic burns a plan quota with a 5-hour and a weekly window, and today the only way to see how close you are is the Z.ai web console. Since the provider list in the status bar is compiled in, there is no setting or plugin that can add it.
Proposed solution
Add Z.ai as a usage provider, following the OpenCode Go shape:
- Settings → AI Provider Accounts → Z.ai: an API key field, stored as a protected secret.
- A fetcher against
GET https://api.z.ai/api/monitor/usage/quota/limit (the endpoint takes the raw key in Authorization — a Bearer prefix is rejected), mapping the returned windows to the 5-hour session and weekly buckets, plus the plan tier (lite / pro / max) as planType.
- A default-on
zai status-bar item with the usual visibility-menu toggle and appearance search entry.
Alternatives considered
Reading the quota from a local file the way Kimi/Grok do — Z.ai keeps no local usage state, so an API read is the only source.
Additional context
Related: #8476 (GLM model support), #10564 (ZCode as a first-class agent), #7569 (custom provider quota visibility). This issue is narrower: usage only, for a plan people already use through the existing agents.
I have a working implementation and will open a PR against this issue.
Problem or use case
Orca tracks usage for Claude, Codex, Gemini, OpenCode Go, Kimi, MiniMax, Antigravity and Grok, but not for Z.ai's GLM Coding Plan. Plenty of people run GLM through Claude Code (
ANTHROPIC_BASE_URL=https://api.z.ai/api/anthropic) or OpenCode; that traffic burns a plan quota with a 5-hour and a weekly window, and today the only way to see how close you are is the Z.ai web console. Since the provider list in the status bar is compiled in, there is no setting or plugin that can add it.Proposed solution
Add Z.ai as a usage provider, following the OpenCode Go shape:
GET https://api.z.ai/api/monitor/usage/quota/limit(the endpoint takes the raw key inAuthorization— aBearerprefix is rejected), mapping the returned windows to the 5-hour session and weekly buckets, plus the plan tier (lite/pro/max) asplanType.zaistatus-bar item with the usual visibility-menu toggle and appearance search entry.Alternatives considered
Reading the quota from a local file the way Kimi/Grok do — Z.ai keeps no local usage state, so an API read is the only source.
Additional context
Related: #8476 (GLM model support), #10564 (ZCode as a first-class agent), #7569 (custom provider quota visibility). This issue is narrower: usage only, for a plan people already use through the existing agents.
I have a working implementation and will open a PR against this issue.