Membership surfaces share a quota pool across web Chat, desktop Agent Chat, and the OpenAI-compatible API.
When you hit limits, the product returns structured errors—not a silent hang. This guide maps what to do next without panic-upgrading blindly.
Prerequisites
- Clavue account when using membership models
- Access to Account → Usage
Steps
- 01
Read the error, not only the toast
Note the code and message (quota exceeded, rate limited, upstream unavailable). Desktop and API should surface a support-friendly
requestIdwhen available—keep it for tickets. - 02
Check Account → Usage
Confirm plan tier, remaining premium/fast pools, and reset times. Boost packs add capacity; they do not replace plan rules.
- 03
Choose the right remediation
- Quota exhausted: upgrade plan, buy a boost pack, or wait for reset.
- Rate limited: slow retries with backoff; avoid parallel spam from scripts.
- Local implementers still work: Claude Code / Codex CLIs use their own auth—switch heavy jobs to terminal panes if membership chat is limited.
- 04
Fix client configuration
For API clients, verify base URL, API key scope, and model product IDs (
auto,clavue-2.1-*). Do not send invented vendor cascade names in public clients.
FAQ
Are consumed turns refundable?
Generally no once quota is spent. See FAQ billing section and plan terms.
Does image gen share chat quota?
Images typically draw from premium-class capacity and cost more than a short chat turn. Prefer text models for pure Q&A.