How to check Codex usage

Where Codex usage actually lives

Codex CLI bills against your OpenAI account, so the authoritative numbers are there, not in any menu-bar app:

• The OpenAI usage dashboard - your account’s usage page shows spend and token consumption over time, broken down by day. This is the source of truth for what a heavy Codex week cost.
• Plan limits - if you’re on a subscription tier with rate limits rather than pure pay-as-you-go, your account settings show the window and where you stand in it.
• Per-run signals in the CLI - Codex surfaces token and context information as it works; keeping an eye on it in the terminal is the fastest read on whether a single long run is getting expensive.

What NapCat shows for Codex - and what it doesn’t

Here’s the honest part. NapCat’s usage meter - the dotted 5-hour and 7-day bars - reads the Claude rate-limit window through your own Anthropic login. It does not read OpenAI usage, so it will not show your Codex tokens or spend. We’d rather tell you that than paint a bar that means nothing.

What NapCat does give you for Codex is the other half of “how much am I using this”: live session activity. Every Codex run is a card in the notch - what it’s doing, which terminal it’s in, whether it’s actively working or has stopped - so you can see how many sessions are going and how long they run. That’s the usage signal a process monitor can honestly report; the token math stays with OpenAI.

A practical way to keep tabs

• Use NapCat’s notch to watch how much you’re running - concurrent Codex sessions and their run lengths, at a glance.
• Use your OpenAI dashboard to watch what it costs - tokens and spend, at the end of the day or week.

More on what Codex support includes is on the Codex page. If you also run Claude Code, NapCat can meter that directly - see how to check Claude Code usage - and Codex notifications on your Mac covers being pinged when a run ends.