Models don't process text character by character — they first split it into tokens. In English a token is about 0.75 words; in Chinese one character is usually 1–2 tokens. Context window sizes and API prices (per million tokens) are both measured in tokens.
When connecting Enclave with your own API key, understanding tokens helps you estimate cost: long conversations and large memory injections both increase input token usage.
Try it yourself in Enclave
Open it in your browser — no credit card, no install.