Observatory / Free Tools / Webpage to LLM Token Counter
Free Webpage to LLM Token Counter & RAG Context Cost Calculator
Measure how many tokens any website’s raw HTML consumes versus stripped prose and /llms.txt—plus exact API grounding costs per 1,000 fetches.
Measure Live HTML vs. Markdown Token Footprint
Token Footprint & Per-1,000-Fetch API Cost: supabase.com
Signal-to-Noise: 0.6%
| Model / Input Pricing Tier | Input Rate / 1M | 1k Reads: Raw HTML (337,794 tok) | 1k Reads: Clean Text (1,905 tok) | 1k Reads: /llms.txt (879 tok) |
|---|---|---|---|---|
| Open-Weight Avg (DeepSeek V4.1 / GLM-5.3) | $0.28 | $94.582 | $0.533 | $0.246 |
| Google Gemini 4 Hydrogen (Fast RAG) | $0.35 | $118.228 | $0.667 | $0.308 |
| Google Gemini 4 Carbon (1M Context) | $1.25 | $422.243 | $2.381 | $1.099 |
| Alibaba Qwen4-Max (1M Multimodal) | $1.50 | $506.691 | $2.858 | $1.319 |
| Frontier Pro Reasoning Tier | $2.50 | $844.485 | $4.763 | $2.198 |
Frequently Asked Questions: Webpage Token Counting & RAG Costs
Technical ReferenceHow many LLM tokens is an average webpage?
Across Sitecube’s empirical benchmark of production websites, raw homepage HTML averages over 130,000 UTF-8 tokens due to inline SVG paths, CSS-in-JS styles, and React/Next.js hydration payloads—while actual human-readable prose averages just 1,500 to 4,500 tokens (often <5% signal-to-noise).
How does /llms.txt reduce LLM token costs?
Serving a structured Markdown manifest at /llms.txt lets coding agents and RAG pipelines bypass noisy DOM markup completely, cutting token consumption by 89.9% to 99.7% per fetch and preventing context-window truncation.