Tokenuze

Tokenuze is a CLI tool that summarizes token and cost usage from various LLM providers. It scans session logs, aggregates token counts, and reports pricing. The output is a table or JSON, suitable for dashboards.

Requirements

• Zig 0.16.0-dev.1456+16fc083f2 or newer (if building from source)
• sqlite3 in your PATH (for Zed and Crush providers)
• Network access to fetch remote pricing / uploading stats (optional)

Supported Providers

• Amp: ~/.local/share/amp/threads
• Claude Code: ~/.claude/projects
• Codex CLI: ~/.codex/sessions
• Crush: .crush/crush.db (project-based, recursively crawled)
• Gemini CLI: ~/.gemini/tmp
• Opencode: ~/.local/share/opencode/storage/session
• Zed: ~/.local/share/zed/threads/threads.db

Installation

• Prebuilt binaries: grab the latest release from the Releases page and put tokenuze on your PATH.
• Build from source: requires Zig 0.16.0-dev.1456+16fc083f2 or newer

git clone https://github.com/bfactory-ai/tokenuze.git
cd tokenuze
zig build --release=fast  # binary will be in zig-out/bin/

Quick Start

tokenuze --upload  # upload usage across all supported models
tokenuze --upload --agent gemini --agent opencode  # request specific agents
tokenuze --since 20250101 --until 20250107  # filter a specific date range
tokenuze --sessions --since 20250101  # print per-session table (default)
tokenuze --sessions --since 20250101 --json --pretty  # print per-session JSON
tokenuze --help  # display all usage options

JSON Output Sample

{
  "daily": [
    {
      "date": "Nov 25, 2025",
      "isoDate": "2025-11-25",
      "inputTokens": 248291670,
      "cachedInputTokens": 236782489,
      "outputTokens": 1188464,
      "reasoningOutputTokens": 749903,
      "totalTokens": 249489543,
      "costUSD": 55.88007429999999,
      "models": {
        "gemini-2.5-flash": {
          "inputTokens": 10992,
          "cachedInputTokens": 0,
          "outputTokens": 10,
          "reasoningOutputTokens": 79,
          "totalTokens": 11032,
          "costUSD": 0.0033225999999999998,
          "pricingAvailable": true,
          "isFallback": false
        },
        "gemini-3-pro-preview": {
          "inputTokens": 5664,
          "cachedInputTokens": 10521,
          "outputTokens": 1738,
          "reasoningOutputTokens": 2048,
          "totalTokens": 16781,
          "costUSD": 0.0342882,
          "pricingAvailable": true,
          "isFallback": false
        },
        "gpt-5.1-codex": {
          "inputTokens": 248275014,
          "cachedInputTokens": 236771968,
          "outputTokens": 1186716,
          "reasoningOutputTokens": 747776,
          "totalTokens": 249461730,
          "costUSD": 55.842463499999994,
          "pricingAvailable": true,
          "isFallback": false
        }
      },
      "missingPricing": []
    }
  ],
  "totals": {
    "inputTokens": 248291670,
    "cachedInputTokens": 236782489,
    "outputTokens": 1188464,
    "reasoningOutputTokens": 749903,
    "totalTokens": 249489543,
    "costUSD": 55.88007429999999,
    "missingPricing": []
  }
}

Missing pricing entries are listed under missing_pricing.

Extending

To add a new provider:

1. Create a new file in src/providers/ and use the provider.makeProvider factory.
2. Implement a parser for the provider's log format that emits model.TokenUsageEvent objects.
3. Use helpers in src/providers/provider.zig for common tasks like JSON parsing and timestamp handling.
4. Register the provider in src/root.zig and expose a loadPricingData function for fallback pricing.
5. Add test fixtures in a new fixtures/<provider> directory and write unit tests in your provider's file.