Estimate Claude tokens. Plan your context.

A private workspace for messages, project instructions, files and output headroom. Local estimates, clearly labeled.

System instructions

CLAUDE.mdTool definitions

or drop them hereTXT, MD, JSON & code · 1 MiB total input limit

Local processing. No upload. No saved prompts.

Your prompt is yours. Text and filenames stay in memory, including when you move to Compare. Only numeric usage is transferred.

View encoding token IDs & accuracy notes

gpt-tokenizer 3.x · o200k_base. Exact for this encoding only. No matching tokenizer has been verified for catalog models. This is not an exact API bill or Claude token count. Chat wrappers, tools and multimodal content can add overhead.

Add text to see token IDs.

Need Claude’s official token count?

Use Anthropic’s token counting endpoint from your own environment. It sends the content to Anthropic; this website does not submit it or collect an API key.

curl https://api.anthropic.com/v1/messages/count_tokens \
  -H "x-api-key: $ANTHROPIC_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "content-type: application/json" \
  -d '{"model":"YOUR_MODEL_ID","messages":[{"role":"user","content":"Hello"}]}'

Official token counting documentation ↗

Count first. Choose with confidence.

Text tokens are only one part of a request. Chat formatting, tool definitions, images and hidden provider overhead can change the billed total.

Read the accuracy notes ↗

What gets shared?

Only numeric counts and workload settings move to the calculator or comparison. Your text and file names never go into the URL.

Privacy by design ↗