Token counter
Count the tokens, then see the monthly bill.
Paste text or upload a .txt or .md file. The count stays in your browser. Pick a model when you want the cost of sending that text to an API.
Your text stays in this browser. It is not sent to CountMyToken.
Words
0
Characters
0
OpenAI tokens
0
OpenAI, exact count
Counted in your browser with OpenAI’s o200k_base encoding, which covers the GPT-5 and GPT-6 models on this site. GPT-4 Turbo and GPT-3.5 use cl100k_base: 0 tokens.
Other models, estimates
Claude, newer tokenizer: 0 tokens
Estimate from Anthropic’s figure of about 2.5 million Unicode characters per million tokens, used from Opus 4.7 onward. This includes Fable, Mythos, Opus 4.7 and later, Sonnet 5 and later, and Haiku 5.5.
Claude, earlier tokenizer: 0 tokens
Estimate from Anthropic’s earlier figure of about 750,000 English words per million tokens, plus one token for each Chinese, Japanese, or Korean character. Used for Opus 4.6 and 4.5, Sonnet 4.6 and 4.5, and Haiku 4.5.
Gemini: 0 tokens
Estimate from Google’s figure of about 4 characters per token.
xAI, Mistral, DeepSeek, and Cohere: 0 tokens
A rule of thumb of about 4 characters per token. These companies have not published a browser tokenizer, so this is not their official count.
How much of the context window is this?
GPT-6 Astra
0 / 1,050,000
GPT-6 Luna
0 / 1,050,000
Claude Opus 5.5
0 / 1,000,000 · estimate
Claude Haiku 5.5
0 / 1,000,000 · estimate
Gemini 4 Argon
0 / 1,048,576 · estimate
Gemini 3.8 Flash
0 / 1,048,576 · estimate
Grok 4.7
0 / 500,000 · estimate
DeepSeek Flash
0 / 1,000,000 · estimate
Models with no confirmed context window are left off the bar.
What 50 short requests a day costs
Each request is 1,000 input tokens and 400 output tokens, sent 50 times a day. Change the numbers.
- GPT-6 Luna
$0.45 a month
- Claude Haiku 5.5
$0.45 a month
- Gemini 2.5 Flash-Lite
$0.39 a month
- DeepSeek Flash
$0.59 a month