Quick Start

Go from zero to your first API response in under 5 minutes. Proxyify is a drop-in replacement for the OpenAI API — no new SDK, no code refactoring, just one line change.

How it works

Proxyify sits between your application and AI providers (OpenAI, Anthropic, Google, Meta, and 50+ others). You send requests to Proxyify using the exact same format as the OpenAI SDK. Proxyify authenticates you, deducts credits, routes your request to the right provider, and returns the response — all transparently.

The only thing that changes in your code is the base_url. Everything else — method names, request parameters, response shapes, streaming — is identical to the OpenAI API.

Prerequisites

Create a Proxyify account

Get an API key

Go to Dashboard → Keys and click New Key. Your key starts with prx- and is shown only once — copy and store it securely. All subsequent uses authenticate via this key.

Check your credits

You can see your balance on the Dashboard overview. Purchase a credit pack from Billing to get started. Credits are deducted per request based on model and tokens used.

Step 1 — Install the SDK

Proxyify uses the standard OpenAI SDK. There is no custom package to install — if you already have openai installed, you are ready to go.

bash — Python

pip install openai

bash — Node.js

npm install openai

Step 2 — Make your first request

Set base_url (Python) or baseURL (JavaScript) to https://proxyify.dev/v1 and pass your prx- key as api_key. That is the only change. All existing OpenAI code works immediately.

Python

from openai import OpenAI

# Initialize the client — only base_url is different from OpenAI
client = OpenAI(
    api_key="prx-xxxxxxxxxxxxxxxx",   # your Proxyify key (prx- prefix)
    base_url="https://proxyify.dev/v1",  # ← the only change
)

# Use exactly like the OpenAI SDK — method names are identical
response = client.chat.completions.create(
    model="openai/gpt-4o-mini",  # provider/model-name format
    messages=[
        {"role": "user", "content": "Explain REST APIs in one sentence."}
    ],
)

print(response.choices[0].message.content)

JavaScript / Node.js

import OpenAI from "openai";

// Initialize the client — only baseURL is different from OpenAI
const client = new OpenAI({
  apiKey:  "prx-xxxxxxxxxxxxxxxx",           // your Proxyify key
  baseURL: "https://proxyify.dev/v1",  // ← the only change
});

// Exact same API as OpenAI
const response = await client.chat.completions.create({
  model:    "openai/gpt-4o-mini",  // provider/model-name format
  messages: [{ role: "user", content: "Explain REST APIs in one sentence." }],
});

console.log(response.choices[0].message.content);

curl

curl https://proxyify.dev/v1/chat/completions \
  -H "Authorization: Bearer prx-xxxxxxxxxxxxxxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "openai/gpt-4o-mini",
    "messages": [{"role": "user", "content": "Explain REST APIs in one sentence."}]
  }'

Step 3 — Read the response

The response structure is identical to the OpenAI API. Proxyify appends a _proxyify object to every response with billing and performance metadata:

json — response

{
  "id": "chatcmpl-...",
  "choices": [{
    "message": { "role": "assistant", "content": "REST APIs are..." },
    "finish_reason": "stop"
  }],
  "usage": { "prompt_tokens": 14, "completion_tokens": 22, "total_tokens": 36 },
  "_proxyify": {
    "credits_used": 4,          // credits deducted from your balance
    "cost_usd": 0.000004,       // actual provider cost in USD
    "model_used": "openai/gpt-4o-mini",
    "input_tokens": 14,
    "output_tokens": 22,
    "latency_ms": 830,          // end-to-end latency
    "cached": false,            // true = served from cache, 0 credits
    "cheaper_model": "anthropic/claude-haiku-3-5"  // null if already cheapest
  }
}

Every request is logged in real time. View full request history, token counts, latency, and cost at Dashboard → Logs.

Model naming

Models are identified by a provider/model-name slug. The provider prefix is required. You can also pass just the model name (gpt-4o) and Proxyify will resolve it automatically — but using the full slug is recommended to avoid ambiguity when multiple providers offer the same model name.

model slugs — examples

# OpenAI
"openai/gpt-4o"
"openai/gpt-4o-mini"
"openai/o3-mini"
"openai/o4-mini"

# Anthropic
"anthropic/claude-3-5-sonnet"
"anthropic/claude-3-7-sonnet"
"anthropic/claude-3-5-haiku"

# Google
"google/gemini-2.0-flash"
"google/gemini-2.5-pro"
"google/gemini-2.5-flash"

# Meta
"meta-llama/llama-4-scout"
"meta-llama/llama-3.3-70b-instruct"

# Image generation
"black-forest-labs/flux-1.1-pro"
"openai/dall-e-3"

# Speech-to-Text — copy slug from Dashboard → Models
"openai/whisper-1"               // example; use exact slug from /dashboard/models/

# Text-to-Speech — copy slug from Dashboard → Models
"openai/gpt-4o-mini-tts"        // example; use exact slug from /dashboard/models/

The full list with pricing is available on the Models page — click any slug to copy it to clipboard. You can also query it programmatically:

curl — list all models

curl https://proxyify.dev/v1/models

Next steps

Authentication → Restrict keys by IP, origin, model, or spending limit. All modalities → Images, video, speech-to-text, text-to-speech — all through one endpoint. Streaming → Enable SSE streaming for real-time token output. Credits & Pricing → How credits are calculated and deducted per request.