Track Anthropic Claude API Costs with Tokonomics Proxy

Swap the Anthropic base URL for Tokonomics to track Claude API costs in real time — your request format and streaming stay exactly the same.

Tokonomics proxies your Anthropic requests without modifying them. You point the Anthropic SDK at https://tokonomics.ca/proxy/anthropic, pass your Tokonomics key as a bearer token, and every Claude call is metered and attributed automatically. Your prompts, message format, and streaming behaviour are unchanged.

The URL change

Replace the Anthropic base URL with your Tokonomics proxy URL. Everything else — your API key, your request bodies, your headers — stays the same.

	URL
Before	https://api.anthropic.com
After	https://tokonomics.ca/proxy/anthropic

A messages request is forwarded to:

https://tokonomics.ca/proxy/anthropic/messages

Python

import anthropic

client = anthropic.Anthropic(
    api_key="your-anthropic-api-key",
    base_url="https://tokonomics.ca/proxy/anthropic",
    default_headers={"Authorization": "Bearer mk_your_tokonomics_key"}
)

message = client.messages.create(
    model="claude-opus-4-5",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Hello!"}]
)

print(message.content[0].text)

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

const client = new Anthropic({
  apiKey: 'your-anthropic-api-key',
  baseURL: 'https://tokonomics.ca/proxy/anthropic',
  defaultHeaders: { 'Authorization': 'Bearer mk_your_tokonomics_key' }
});

const message = await client.messages.create({
  model: 'claude-opus-4-5',
  max_tokens: 1024,
  messages: [{ role: 'user', content: 'Hello!' }]
});

console.log(message.content[0].text);

Cache token tracking

Tokonomics automatically captures Anthropic’s prompt caching token fields from every response. Alongside standard input_tokens and output_tokens, your usage records include:

• cache_read_tokens — tokens served from the prompt cache (billed at the reduced cache-read rate)
• cache_creation_tokens — tokens written into the prompt cache (billed at the cache-write rate)

You can view cache savings on the Usage dashboard. No additional configuration is needed — Tokonomics reads these fields from every response that includes them.

Tagging works the same as OpenAI. Add an X-Metering-Tags header to any request — for example, {"team":"platform","feature":"document-qa"} — to attribute costs to teams or features. See Tagging for details.