Quick Start

This page provides the minimum startup flow to get a working proxy instance quickly.

1. Prepare dependencies

Download the binary for your platform from release.

2. Prepare config file

You can start from the full example config.

Minimal working example:

[global]
host = "0.0.0.0"
port = 8787
proxy = ""
admin_key = "replace-with-strong-admin-key"
mask_sensitive_info = true
data_dir = "./data"
dsn = "sqlite://./data/gproxy.db?mode=rwc"

[[channels]]
id = "openai"
enabled = true

[channels.settings]
base_url = "https://api.openai.com"

[[channels.credentials]]
secret = "sk-replace-me"

3. Start service

Run the downloaded binary directly:

# Linux / macOS
./gproxy

# Windows
gproxy.exe

After startup, logs should show:

• Listen address (default http://0.0.0.0:8787)
• Current admin key (password:)

If gproxy.toml is missing, the service starts with in-memory defaults and auto-generates a 16-char admin key.

4. Send a minimal verification request

curl -sS http://127.0.0.1:8787/openai/v1/models \
  -H "x-api-key: <your user key or admin key>"

If you get a model list JSON, routing, auth, and upstream connectivity are all working.

5. Open admin console

• Console entry: GET /
• Static assets path: /assets/*