Local-first Codex analytics

See where your
Codex tokens go.

Run one command to turn your local Codex logs into a private usage dashboard—tokens, model mix, rate-limit snapshots, and API-equivalent cost.

Connect my Codex →View sample dashboard
✓ No prompts✓ No source code✓ No file contents
codexmeter / overview Synced 2m ago

YOUR CODEX USAGE

Good morning, Alex.

Total tokens

13.37M↗ 18.4%

API-equivalent

$25.41estimate

Cache hit rate

78.2%↗ 6.1%

DAILY TOKEN USE

Input + output + cached input
July 1–30
Jul 1Jul 15Jul 30

5-hour window

63% leftResets in 2h 14m

Weekly window

82% leftResets Friday

One private view for the numbers Codex leaves on your machine.

13.37M tokens tracked
78% cache hit rate
2 platforms supported
CONNECT IN MINUTES

One command. Your usage, explained.

Follow the steps below. Keep this page open and your dashboard will appear after the first sync.

  1. 1
    Sign in with GitHub

    Create your private dashboard and a one-time connection code.

  2. 2
    Choose your system

    We tailor the command for macOS or Windows PowerShell.

  3. 3
    Copy and run one command

    The collector previews, filters, and syncs usage metadata only.

  4. 4
    See your dashboard

    Your first report usually appears in under a minute.

PREVIEW MODE GitHub OAuth and the CLI package are awaiting production credentials.
HOW IT WORKS

Useful numbers leave your computer.
Your work does not.

01
⌁

Read locally

The collector finds active and archived Codex session logs on your Mac or Windows PC.

02
≋

Strip content

Prompts, responses, tool output, code, file paths, project names, skills, and MCP names are discarded.

03
↑

Sync totals

Only aggregated token counts, model names, timestamps, and optional rate-limit snapshots are uploaded.

04
▦

Explain usage

Your dashboard turns those totals into trends, cache metrics, and an API-equivalent cost estimate.

PRIVACY BY DESIGN

Your code is none
of our business.

The collector is designed around a narrow data contract. You can preview the exact payload before the first upload.

See the connection flow →

What is synced

✓ Aggregated token counts

✓ Model names

✓ Dates and timezone

✓ Optional limit snapshots

✓ Anonymous device ID

What never leaves

× Prompts and responses

× Source code or files

× Tool calls and commands

× Project or repository names

× Skill and MCP names

COST CONTEXT

Tokens are easier to understand when they have a price.

We apply effective-dated public model prices to uncached input, cached input, and output tokens. Estimates are educational—not your subscription bill.

Track my usage →
MODELTOKENSEST. COST
GPT-5.6 Sol
8.42M$18.73
GPT-5.4 mini
3.21M$4.18
Other Codex models
1.74M$2.50
i API-equivalent estimate · Pricing version Jul 2026
READY WHEN YOU ARE

Know your Codex usage
before it surprises you.

Connect my Codex →