HomeProvidersConfigurationCLI Reference

Providers

Supported AI Providers

LimitAID queries three AI platforms, each with its own rate limiting and billing model.

1

Codex / OpenAI

Codex / OpenAI

Hits the chatgpt.com usage API and returns two independent rate-limit windows: a rolling 5-hour window for short-burst work, and a weekly window for sustained usage. It also reports your credits balance and any Spark sub-limits that apply to your plan tier. Credentials are pulled automatically from ~/.codex/auth.json.

Primary window5-hour rolling
Secondary windowWeekly reset
CreditsBalance reported
Spark sub-limitsPer-tier caps
API: GET https://chatgpt.com/backend-api/wham/usage
Auto-discovery: Reads ~/.codex/auth.json
bash
2

OpenRouter

OpenRouter

Exposes a single key endpoint that returns your credit limit, remaining balance, and a usage breakdown across daily, weekly, and monthly windows. Unlike Codex, there's no auto-discovery — you need to add your API key to keys.conf by hand.

Credit limitConfigured per key
RemainingBalance + percentage
Usage windowsDaily / Weekly / Monthly
Free tierFlagged in response
API: GET https://openrouter.ai/api/v1/key
Configuration: Manual — add to keys.conf
bash
3

Z.ai

Z.ai

Reports two hard limits: a 5-hour time-limit window that gates how long you can run agents, and a weekly token budget. The per-tool MCP breakdown lets you see exactly which tools (search-prime, web-reader, zread, etc.) are eating into your quota. Credentials come from your opencode config.

Time limit5-hour window
Token budgetWeekly cap
MCP breakdownPer-tool counters
Auto-discoveryopencode config
API: GET https://api.z.ai/api/monitor/usage/quota/limit
Auto-discovery: Reads ~/.config/opencode/opencode.json
bash