---
title: "LLM Provider Accounts"
id: "1716"
type: "page"
slug: "accounts"
published_at: "2026-09-20T20:36:25+00:00"
modified_at: "2026-09-21T01:36:28+00:00"
url: "https://xedant.com/agents/multiagent/docs/accounts"
markdown_url: "https://xedant.com/agents/multiagent/docs/accounts.md"
excerpt: "The “Accounts” section tracks your LLM provider accounts and watches their quotas and balances while…"
---

# LLM Provider Accounts

[https://xedant.com/agents/multiagent/docs/accounts.md](https://xedant.com/agents/multiagent/docs/accounts.md)

The “Accounts” section tracks your LLM provider accounts and watches their quotas and balances while your agents work. When an account runs out of quota or money, you hear it from MultiAgent — not from crashing chats.

## Account tracking

Supported providers are Z.ai, BigModel (Z.ai’s Chinese endpoint), Codex, DeepSeek and OpenRouter. Each account records a provider, credentials, a status, colored tags and filters for them — so you can group “team A’s work accounts” or “the overnight reserve”. Counters at the top show how many accounts are at risk and how many are being watched.

## Quota and balance monitoring

A background poller — a small watcher inside MultiAgent — checks the providers on a schedule and records state snapshots: the short-term history is kept as snapshots, the long-term as hourly rollups. That feeds the charts: balance over time, quota over time, balances and quotas per provider and per account, distributions and patterns by hour and day of week, plus a snapshot table. Quota updates reach the UI live — charts and badges refresh without reloading.

## Signing in to Codex

Codex accounts can sign in with ChatGPT right from MultiAgent — a device login over HTTP: the “Sign in with ChatGPT” button starts the flow, you can check its status or cancel, and signing out clears the link. You never have to copy the code to OpenAI’s page by hand — MultiAgent walks the flow from start to finish.

## Codex quotas

Codex reports two limit windows: 5 hours and weekly. MultiAgent shows how full each window is and when it resets next, warns when a reset is close, and takes peak hours into account. When you plan a heavy task, you can see whether it fits the window.

## Models and limits

Models connect accounts to agents: attach a model to an agent and its keys land automatically in the container’s environment variables. Per model you set the limit on parallel chats and the delay between calls — the same mechanism the [prompt queue](/agents/multiagent/docs/chats)
 and scheduled tasks use when starting new chats.

[← Back to documentation](/agents/multiagent/docs)