Codex usage tracking on ChatGPT plans.
ChatGPT plans have no usage API, but Codex can export its own usage over OpenTelemetry. Point that export at Audon with an ingest token and token usage per day and model arrives while you work.
What Audon reads. And what it never reads.
Reads
- Tokens per day and model (input, cached input, output, reasoning)
- Turns
- Codex’s cost estimate when it reports one (API-key sign-in only)
Never reads
- Prompts, responses, code or tool output
- Billed spend or the plan fee (add the plan under Subscriptions)
- Usage from before the setup
The key to create. Used read-only, encrypted, never shown again.
- Credential
- No provider key. Codex sends its own OpenTelemetry metrics to Audon with an Audon ingest token (aud_otlp_…).
- Where to create it
- For Codex signed in with ChatGPT (Plus, Pro, Business). Create the connection with connection_create_otlp (tool "codex"), then add the returned [otel] block to ~/.codex/config.toml and restart Codex.
- Provider documentation
- developers.openai.com/codex/config-advanced
- Step by step
- Provider setup in the docs
Good to know.
- ChatGPT plans report tokens only, so nothing becomes spend. Add the plan fee under Settings → Subscriptions.
- Usage from before the setup isn’t included.
Questions about Codex.
What does Audon need to connect Codex?
No provider key. Codex sends its own OpenTelemetry metrics to Audon with an Audon ingest token (aud_otlp_…).. For Codex signed in with ChatGPT (Plus, Pro, Business). Create the connection with connection_create_otlp (tool "codex"), then add the returned [otel] block to ~/.codex/config.toml and restart Codex.
What does Audon read from Codex?
Tokens per day and model (input, cached input, output, reasoning), Turns and Codex’s cost estimate when it reports one (API-key sign-in only).
Does Audon read prompts, responses or code?
No. Audon doesn’t read: prompts, responses, code or tool output, billed spend or the plan fee (add the plan under Subscriptions) and usage from before the setup.
How do I stop it?
Disconnect in Integrations: the ingest token stops working at once. Remove the telemetry settings from the tool as well, so it stops sending.
Last updated
Other integrations.
- What we read
OpenAI
Costs and usage by project, model and line item, read with an organization Admin key.
- What we read
Anthropic (Claude API)
Cost and usage reports by workspace and model, plus Claude Code per person, with one Admin key.
- What we read
Claude EnterpriseNew
Per-person activity, idle seats, and daily cost and tokens by product and model.
- What we read
Claude CodeNew
Tokens and cost per person and model, from the Admin report or Claude Code’s own telemetry.
- What we read
xAINew
Daily team spend by model and endpoint, read with a team-scoped management key.
- What we read
CursorNew
Members, usage per person and model and on-demand spend, or the usage export on individual plans.
- What we read
DevinNew
Daily ACUs by product and per person, priced with your own price per ACU.