AI infrastructure for developers

One API, meteredSpeaks your SDKs

Call available models through one compatible endpoint, with usage deducted from your prepaid USD balance at published token rates. Top up from US$10; balance never expires.

Get started View docs

USD: USD balance
US$10: minimum top-up
Never: balance expiry (never)
1M: context window (tokens)

Integration examplequickstart.py

from openai import OpenAI

client = OpenAI(
    base_url="https://proxystacks.com/v1",
    api_key="YOUR_RELAY_API_KEY",
)

response = client.chat.completions.create(
    model="YOUR_MODEL_ID",
    messages=[
        {"role": "user", "content": "Hello!"}
    ],
)

Example code · replace the key and use an available model ID from your console

Built around familiar API protocols

Chat Completions

/v1/chat/completions

Use the familiar messages request format.

Responses API

/v1/responses

Send model requests with the Responses format.

Anthropic Messages

/anthropic/v1/messages

Connect an Anthropic-compatible client.

Quickstart

From API key to first request in three steps.

Create an API key

Create a key in your console. Store the full value securely; it is shown only once.

POST /v1/keys

Configure your client

Set your Relay base URL and key, then choose an available model in the console.

BASE_URL="https://proxystacks.com/v1"

Send a model request

Send requests with a compatible SDK. Usage is deducted from your USD balance at published token rates.

model: "deepseek-v4-1-flash"

Platform

One integration. Flexible usage.

One integration point: Access configured model routes through compatible protocols, with fewer separate integrations to maintain.
Centralized API keys: Create and revoke Relay keys from the console to keep application access and credentials organized.
Flexible top-ups: Usage is deducted from your USD balance at published token rates—usage, cost, and balance are visible in real time.

Ready to send your first request?

Create a key and connect your application.

Get started