LLM Proxy

Base URL: https://api.nozle.app/api/v1 Reverse proxy routes that forward LLM requests to OpenAI or Anthropic and automatically track token usage. The proxy reads the response to extract { model, input_tokens, output_tokens, latency_ms } and fires an internal nozle.track() event — no SDK wrapper needed on your side.

LLM proxy routes use X-Nozle-Key for Nozle authentication, not the Authorization header. The Authorization header is reserved for the customer’s own LLM provider API key, which the proxy forwards untouched to the provider.

/api/v1/proxy/openai/v1/*

Proxies all requests to https://api.openai.com/v1/*. Any OpenAI-compatible endpoint works (chat completions, embeddings, etc.). Auth: X-Nozle-Key header (sk_ only) Required headers:

Header	Description
`X-Nozle-Key`	Your Nozle secret key (`sk_live_...`)
`X-Nozle-Customer`	External customer ID to bill for this usage
`Authorization`	Customer’s OpenAI API key (`Bearer sk-...`)

cURL

curl -X POST https://api.nozle.app/api/v1/proxy/openai/v1/chat/completions \
  -H "X-Nozle-Key: sk_live_your_nozle_key" \
  -H "X-Nozle-Customer: cust_123" \
  -H "Authorization: Bearer sk-customer-openai-key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "<openai-model>",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

/api/v1/proxy/anthropic/v1/*

Proxies all requests to https://api.anthropic.com/v1/*. Auth: X-Nozle-Key header (sk_ only) Required headers:

Header	Description
`X-Nozle-Key`	Your Nozle secret key (`sk_live_...`)
`X-Nozle-Customer`	External customer ID to bill for this usage
`Authorization`	Customer’s Anthropic API key

cURL

curl -X POST https://api.nozle.app/api/v1/proxy/anthropic/v1/messages \
  -H "X-Nozle-Key: sk_live_your_nozle_key" \
  -H "X-Nozle-Customer: cust_123" \
  -H "x-api-key: customer-anthropic-key" \
  -H "anthropic-version: 2023-06-01" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "<anthropic-model>",
    "max_tokens": 1024,
    "messages": [{"role": "user", "content": "Hello"}]
  }'

The proxy returns the provider’s response unchanged. Your customer’s API key is forwarded but never stored. Token usage is extracted from the response and tracked automatically as a billing event on the llm_tokens metric.

Overview

Application Workflows

Billing Resources

Usage & Metering

Invoicing & Payments

Platform & Operations

/api/v1/proxy/openai/v1/*

/api/v1/proxy/anthropic/v1/*

​/api/v1/proxy/openai/v1/*

​/api/v1/proxy/anthropic/v1/*

/api/v1/proxy/openai/v1/*

/api/v1/proxy/anthropic/v1/*