A CLI tool for tracking and analyzing OpenAI Codex CLI token usage by repository, date, and model.
- 📊 Repository Breakdown: See which projects consume the most tokens
- 📅 Daily Reports: Track usage over time with daily aggregations
- 🤖 Model Distribution: Understand which Codex models you're using
- 💰 Cost Tracking: Calculate estimated costs based on model pricing
Costs are estimated based on OpenAI's Standard tier pricing. Actual costs may vary based on:
- Pricing tier (Batch, Flex, Standard, Priority)
- Volume discounts
- Enterprise agreements
- Currency conversion
- Special Codex pricing tiers
For the most accurate costs, check your OpenAI billing dashboard or use npx @ccusage/codex@latest which fetches live pricing from LiteLLM.
- 🎨 Beautiful Tables: Colorful, formatted terminal output
- 📄 JSON Export: Get structured data for further analysis
# Install globally
npm install -g codex-usage
# Or run with npx
npx codex-usageShows token usage grouped by repository/folder:
codex-usage repos
codex-usage repos --top 10 # Show only top 10 repos
codex-usage repos --since 2026-01-01 # Filter by date
codex-usage repos --json # JSON outputShows daily token usage:
codex-usage daily
codex-usage daily --since 7days # Last 7 days
codex-usage daily --until 2025-12-31 # Until specific dateShows usage grouped by Codex model:
codex-usage modelsShows overall usage summary:
codex-usage summary--codex-dir <path>: Codex data directory (default:~/.codex)--json: Output as JSON--since <date>: Start date (YYYY-MM-DD or relative: 7days, 1month)--until <date>: End date (YYYY-MM-DD)--top <n>: Limit results (repos command)
╔════════════════════════════════════════════════════════════╗
║ Codex Token Usage by Repository ║
╚════════════════════════════════════════════════════════════╝
Sep 15, 2025 - Jan 10, 2026
┌──────────────────────────────────────────────────────┬───────────────┬───────────┬─────────┐
│ Repository │ Total Tokens │ Cost │ % Total │
├──────────────────────────────────────────────────────┼───────────────┼───────────┼─────────┤
│ ~/Projects/acme/alpha-app │ 821,324,637 │ $654.35 │ 25.2% │
│ ~/Projects/acme/beta-service │ 688,790,294 │ $555.88 │ 21.1% │
│ ~/Projects/acme/gamma-tool │ 399,651,942 │ $322.35 │ 12.3% │
├──────────────────────────────────────────────────────┼───────────────┼───────────┼─────────┤
│ TOTAL │ 3,259,607,066 │ $2,614.47 │ 100% │
└──────────────────────────────────────────────────────┴───────────────┴───────────┴─────────┘
🤖 Model Distribution
──────────────────────────────────────────────────
gpt-5.2-codex 2,123,184,604 65.1%
gpt-5.1-codex-max 432,212,433 13.3%
gpt-5-codex 310,992,712 9.5%
The tool uses current OpenAI Codex pricing:
- GPT-5.1 Codex: $0.0015/1K input, $0.0060/1K output
- GPT-5.0 Codex: $0.0010/1K input, $0.0040/1K output
- Cache reads are discounted at 50%
Pricing is hardcoded for reliability. For the latest pricing, visit OpenAI's pricing page.
The tool reads Codex session data from ~/.codex/sessions/ (or custom path with --codex-dir).
Each session file contains JSONL records with:
- Session metadata (cwd, model, git info)
- Turn context (working directory, model changes)
- Token usage events with detailed breakdowns
# Install dependencies
npm install
# Build
npm run build
# Run locally
npm start -- repos
# Development mode (watch)
npm run devMIT
Contributions are welcome! Please open an issue or PR.