Internal example: a session’s project cost

One session, one project

This internal example follows recorded Codex tokens into a project total. The account, device, repository and project are synthetic; it is not a customer case study.

The session

On 14 September 2026, Codex CLI 0.154.0 on macOS reported gpt-6-astra: 18,520 input tokens, including 11,904 cached, plus 10 output tokens. It returned a fixed marker without tools or file reads. Input includes CLI context; reasoning and cache writes were zero.

Passcard’s frozen 2026-09-10 snapshot uses gpt-5.6 fallback rates for the gpt-6 family: $5/$0.50/$30 per million uncached/cached/output tokens. These are unverified for gpt-6-astra.

6,616 × $5 ÷ 1,000,000 + 11,904 × $0.50 ÷ 1,000,000 + 10 × $30 ÷ 1,000,000 = $0.039332. Download the sanitized counters and rates.

Assign the repository

In the dashboard’s project inbox, map the session’s repository to a project. This example maps github.com/example/cost-walkthrough to Cost walkthrough. Compare the same date range; a project may contain other sessions.

What passed

A local PostgreSQL test used the recorded counters and a synthetic session envelope with Passcard’s production ingestion and reporting code:

• First upload: one event accepted, cost flagged as estimated.
• Dashboard, project and model totals: $0.039332, in both whole-day and partial-window queries.
• Replay: zero accepted, one duplicate, total unchanged.

The display

This verifies a local calculation and project assignment. Live authentication, device pairing and cloud delivery were not tested. It establishes neither a provider charge nor customer savings.

Try it

Follow the Codex setup guide, map one repository, then compare its counts and dated rates.