LLM Token Counter and Cost Estimator
Count the tokens in whatever you paste and see, as a bar, how much of the selected model's context window it fills. Prices come from the table LiteLLM maintains, so nothing here is a hand-copied rate.
0Tokens
0 chars · 0.00 tokens per char
Model
61 shown of 295
- Provider
- OpenAI
- Input price
- $2.5 / 1M tokens
- Output price
- $10 / 1M tokens
- Context
- 128,000
- Max output
- 16,384
- Input
- $0.00000
- Output
- $0.00500
- Per run
- $0.00500
- 1×
- $0.00500
What you enter never leaves your device. Prices and context windows come from LiteLLM's table (MIT), fetched 2026-08-14. They vary by account and region — check the provider's own pricing before committing real money. Token counts use OpenAI's o200k, so Claude and Gemini figures are approximate.
Frequently asked questions
Is the count accurate for Claude or Gemini?
It is an approximation. The tokenizer used here is OpenAI's o200k (GPT-4o and later), and Claude and Gemini each use their own. The result usually lands within 10–20%, but do not rely on it when you are cutting a context limit close.
Where do the prices come from?
From model_prices_and_context_window.json (MIT), maintained by LiteLLM. It is the de facto shared source that many tools pull from, and far better than a table we would maintain ourselves — one we wrote would start going stale the same day. The fetch date is shown on the page. Rates can vary by account and region, so check the provider's own pricing before committing to a large spend.
Why do some languages use more tokens than others?
Tokenizers are trained mostly on English text, so English averages around 0.25 tokens per character while Korean runs 0.4 to 0.6. The same content can cost twice as much to send in one language as another. Tokens per character are shown so you can see the ratio for your own text.
Does anything I paste get sent to a server?
No. Everything runs inside your browser and nothing is uploaded or stored. That is why you can use it on things you would not normally paste into a web page, like an internal config file or a production query. It also works with your network disconnected.