Usage

Arctic aggregates usage, costs, and (when available) provider limits in one place.

Command

arctic run --command usage

TUI shortcut

In the TUI prompt:

/usage

This opens the Usage dialog with detailed information for all providers.

To manage multiple connections:

/connections

This opens the Manage Connections dialog.

Sidebar Usage Display

Usage limits are also displayed in the sidebar (bottom section) for supported providers. This gives you quick visibility into your current usage without opening the full dialog.

Multiple connections per provider

You can connect multiple accounts to the same provider (for example, two Codex accounts) and track usage limits separately for each connection. Use arctic auth login in the CLI or /connect in the TUI, choose your provider, and you will be prompted to add another connection when one already exists. Use /connections to view and manage named connections.

Supported providers:

• Codex
• Z.ai coding plan
• Anthropic (Claude)
• Google (Gemini)
• GitHub Copilot
• Kimi for Coding
• QwenCode
• MiniMax

What you’ll see

• Provider tabs: switch providers with Tab
• Access status: allowed / blocked / limit reached
• Plan (if available)
• Updated time
• Token usage: total/input/output/cached
• Cost summary: total + breakdown (if available)
• Limits: primary/secondary windows with % remaining + reset time (when available)

How limits work

Usage data comes from two sources:

1. Session usage (tokens + cost)
   Arctic computes this from the local message log and model pricing. This is available for all providers that report token counts in messages and have pricing data (via models.dev or provider pricing).

2. Provider limits (quota/rate limits)
   Only some providers expose account limits via an API. Arctic currently fetches limits for:

   • Codex
   • Z.ai coding plan
   • Anthropic (Claude subscription)
   • GitHub Copilot
   • Google (Gemini Code Assist)
   • Kimi for Coding
   • QwenCode
   • MiniMax

If a provider doesn’t expose limits, you’ll still see session token usage and cost, but the limits section may be empty.