{
  "measured_at": "2026-06-23",
  "tiktoken_version": "0.13.0",
  "sample_topic": "What is JSON? introduction with h2 + paragraphs + bullet list + code block + 3-column table",
  "samples": {
    "html": {
      "file": "sample.html",
      "chars": 2911,
      "bytes_utf8": 2911,
      "tokens_cl100k_base": 832,
      "tokens_o200k_base": 835,
      "chars_per_cl100k_token": 3.50
    },
    "markdown": {
      "file": "sample.md",
      "format_note": "CommonMark + GFM table extension",
      "chars": 1071,
      "bytes_utf8": 1071,
      "tokens_cl100k_base": 243,
      "tokens_o200k_base": 247,
      "chars_per_cl100k_token": 4.41
    },
    "plain_text": {
      "file": "sample.txt",
      "chars": 986,
      "bytes_utf8": 986,
      "tokens_cl100k_base": 213,
      "tokens_o200k_base": 217,
      "chars_per_cl100k_token": 4.63
    }
  },
  "reduction_vs_html": {
    "markdown": {
      "chars_pct": -63.2,
      "tokens_cl100k_pct": -70.8,
      "tokens_o200k_pct": -70.4
    },
    "plain_text": {
      "chars_pct": -66.1,
      "tokens_cl100k_pct": -74.4,
      "tokens_o200k_pct": -74.0
    }
  },
  "reproduction_steps": [
    "tmp/markdown-vs-html-benchmark/ 内の sample.{html,md,txt} を見る",
    "python3 -m venv tmp/markdown-vs-html-benchmark/venv",
    "tmp/markdown-vs-html-benchmark/venv/bin/pip install tiktoken",
    "tmp/markdown-vs-html-benchmark/venv/bin/python tmp/markdown-vs-html-benchmark/measure.py"
  ]
}
