Count tokens

POST/v1/messages/count_tokens

Count the input tokens of a Messages request: Anthropic's own count for an Anthropic model, otherwise an estimate marked as one.

Open Memtro

Counts the input tokens of a Messages request, so an SDK that budgets before it asks keeps working. Where the model is an anthropic/* model and you hold an Anthropic key, the count is Anthropic's own; otherwise, or if that call fails or takes over ten seconds, it is an estimate of about 3.6 characters a token, marked memtro.estimated.

It counts your request, not what Memtro sends upstream: the prompt, tools and recalled context Memtro adds are not yours to budget for, and they vary with where the request routes.

Example

Against https://platform.memtro.com, with your key in MEMTRO_API_KEY.

curl https://platform.memtro.com/v1/messages/count_tokens \
  -H "x-api-key: $MEMTRO_API_KEY" \
  -H "content-type: application/json" \
  -d '{
    "model": "anthropic/claude-opus-5",
    "messages": [{"role": "user", "content": "What changed on the Baltic account this week?"}]
  }'

from anthropic import Anthropic

client = Anthropic(base_url="https://platform.memtro.com", api_key="sk-memtro-…")

count = client.messages.count_tokens(
    model="anthropic/claude-opus-5",
    messages=[{"role": "user", "content": "What changed on the Baltic account this week?"}],
)
print(count.input_tokens)

import Anthropic from "@anthropic-ai/sdk";

const client = new Anthropic({ baseURL: "https://platform.memtro.com", apiKey: process.env.MEMTRO_API_KEY });

const count = await client.messages.countTokens({
  model: "anthropic/claude-opus-5",
  messages: [{ role: "user", content: "What changed on the Baltic account this week?" }],
});
console.log(count.input_tokens);

A model with no Anthropic tokeniser behind it gets an estimate, marked as one.

{ "input_tokens": 18, "memtro": { "estimated": true } }

Authentication

An API key, as x-api-key or Authorization: Bearer.

This route sets no request id header.
Browsers may call it from any origin: it answers CORS preflights and allows the Authorization, x-api-key and Anthropic version headers.

Request body

The Messages request without max_tokens. The fields that matter to the count:

modelstringRequired: The model the request is bound for.
messagesarray of objectsRequired: As on Messages.
systemstring or array of objectsOptional: The system prompt.
toolsarray of objectsOptional: Your tools; their schemas count.
tool_choiceobjectOptional: As on Messages.

Response

200application/json

The count.

input_tokensintegerOptional: The number of input tokens.
memtroobjectOptional: {"estimated": true} when the number is an estimate rather than Anthropic's count.

Errors

In Anthropic's envelope, {"type": "error", "error": {"type", "message"}}, with no code: the type is chosen from the status. See Errors on the overview.

Status	Type	When
400	`invalid_request_error`	The body failed validation.
401	`authentication_error`	The key is missing, malformed, revoked or expired.
402	`authentication_error`	The organisation has no active subscription. This route types every refusal of the key that way.

Messages