Global usage
Codex credits are not USD. Effort affects tokens used per request, not the per-token credit rate.
Tokens over time
Per-profile capacity & usage
Codex accounts
Each account is a separate ChatGPT login (your seat). Requests are routed by the shares below; accounts over their threshold are skipped and requests fail over automatically. OpenRouter accounts are API keys and route to openrouter.ai instead of ChatGPT.
Groups & routing shares
Ungrouped accounts make up the Default group. Create groups to route specific API keys to specific accounts; an account can belong to several groups. Each group splits its own traffic — drag the handles below. Accounts on cooldown are skipped and the rest re-balance.
Accounts
API keys
Give these keys to client software (like gbrain). Point the client's OpenAI base URL at
/v1 and use one of these keys as the API key. A key routes
only to accounts in its assigned group (Default = ungrouped accounts).
Request logs
| Time | Endpoint | Model | Codex account | Group | Client | Effort | Tok (total/reason) | Codex credits | Dropped params | Dropped params (values) | Status | Latency | Tries | Error |
|---|
Settings
When an account uses this much of any rate-limit window, it's put on cooldown and traffic routes elsewhere. Accounts can override this individually on the Accounts tab.