API reference

Count tokens

POST /v1/messages/count_tokens estimates how many input tokens a Messages request would use. It costs nothing.

POSThttps://gateway.kalsenti.ai/v1/messages/count_tokens
On this page

Request

Send the body you’d send to Messages, without max_tokens. The same headers are needed, including anthropic-version. The Python and JavaScript snippets use an Anthropic SDK client set up as on the SDKs page.

curl https://gateway.kalsenti.ai/v1/messages/count_tokens \
  -H "x-api-key: $KALSENTI_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-v4.1-flash-1017-raw",
    "system": "You are a concise assistant.",
    "messages": [{ "role": "user", "content": "Name three sea birds." }]
  }'

Response

Response
{ "input_tokens": 40 }

An unknown model gets a 404 with not_found_error, and a body that can’t be read as a Messages request gets a 400, as it would from Messages itself.

How the count is made

The count is an estimate. The tokens you’re charged for can be a little different.

  • It’s worked out from the size of the request, at about four bytes per token, because every model counts tokens a little differently.
  • It doesn’t reach a model and costs nothing, and it doesn’t appear on the Usage page.
  • It does count as a request towards your key’s rate limit.