Build Your Own AI Proxy Server: Master sub2api & Docker Deployment for Cost-Effective GPT-5.5 Access

Tutorial: How to Build a self-hosted AI Proxy Server

💡 Core Concept: We are using an open-source engine called sub2api. It ACTs as a "translator" and "traffic dispatcher," converting the web session capability of your chatgpt accounts into OpenAI/anthropic protocol-compliant API Keys.

1. Prerequisites & Environment Setup

• Server: A Virtual Private Server (VPS) located overseas (e.g., AWS, Tencent Cloud, etc.). I recommend a 2 Core 4GB configuration running Ubuntu (e.g., Tencent Cloud at ~$199/year).

• Location: Select nodes in Japan, USA (Silicon Valley), or Europe. Avoid Hong Kong nodes for stability.

• Access: Root SSH access to the server.

ssh root@your_server_ip

Step 2: Update System & Install Dependencies
Update the package list and install basic dependencies:

sudo apt update && sudo apt upgrade -y
sudo apt install -y ca-certificates curl gnupg

Step 3: Install Docker & docker Compose
sub2api relies heavily on Docker. Add the official Docker repository and install the engine:

# Add Docker's official GPG key
sudo install -m 0755 -d /etc/apt/keyrings
sudo curl -fsSL https://download.docker.com/Linux/ubuntu/gpg | sudo gpg --deArmor -o /etc/apt/keyrings/docker.gpg
sudo chmod a+r /etc/apt/keyrings/docker.gpg

# Set up the repository
echo \
  "deb [arch=$(dpkg --print-architecture) signed-by=/etc/apt/keyrings/docker.gpg] https://download.docker.com/linUX/ubuntu \
  $(. /etc/os-release && echo "$VERSION_CODENAME") stable" | \
  sudo tee /etc/apt/sources.list.d/docker.list > /dev/null

# Install Docker Engine
sudo apt update
sudo apt install -y docker-ce docker-ce-CLI containerd.io docker-buildx-plugin docker-compose-plugin

# Start and enable Docker service
sudo systemctl enable --now docker

Step 4: Verify installation
Run a test container to ensure Docker is working:

sudo docker run hello-world

2. Deploy sub2api Service

mkdir -p sub2api
cd sub2api
curl -sSL https://raw.GitHubusercontent.com/Wei-Shaw/sub2api/main/deploy/docker-deploy.sh | bash

Step 2: Start the Container

docker compose up -d

• If the page is unreachable, check the Docker service status: docker compose ps.

• Ensure your server's Firewall and Security Group rules allow traffic on port 8080.

docker compose logs sub2api | grep "admin password"

• Username: admin@sub2api.local

• Password: (The one-time password shown in the logs)

3. Secure the Server with SSL (HTTPS)

curl https://get.acme.sh | sh

# Stop Nginx to free port 80
systemctl stop nginx

# Issue certificate (Replace with your domain)
/root/.acme.sh/acme.sh --issue -d sub2api.yourdomain.com --standalone --force

Step 4: Install Certificate to Target Directory
Copy the certificate files to your sub2api SSL folder:

# Create SSL directory
mkdir -p /root/sub2api/ssl

# Install certificate
/root/.acme.sh/acme.sh --install-cert -d sub2api.yourdomain.com \
--key-file       /root/sub2api/ssl/sub2api.key  \
--fullchain-file /root/sub2api/ssl/sub2api.pem \
--reloadcmd     "systemctl reload nginx"

Step 5: Configure Nginx reverse proxy
Create an Nginx configuration file (/etc/nginx/conf.d/sub2api.conf) with the following content:

server {
    listen 80;
    server_name sub2api.yourdomain.com;
    return 301 https://$host$request_uri;
}

server {
    listen 443 ssl;
    server_name sub2api.yourdomain.com;

    underscores_in_headers on;

    # SSL Certificates
    ssl_certificate /root/sub2api/ssl/sub2api.pem;
    ssl_certificate_key /root/sub2api/ssl/sub2api.key;

    ssl_protocols TLSv1.2 TLSv1.3;
    ssl_ciphers ECDHE-ECDSA-AES128-GCM-SHA256:ECDHE-RSA-AES128-GCM-SHA256:ECDHE-ECDSA-AES256-GCM-SHA384:ECDHE-RSA-AES256-GCM-SHA384;
    ssl_prefer_server_ciphers on;

    # Reverse Proxy to sub2api
    location / {
        proxy_pass http://127.0.0.1:8080;
        proxy_set_header Host $host;
        proxy_set_header X-Real-IP $rEMOte_addr;
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
        proxy_set_header X-Forwarded-Proto $scheme;
        proxy_http_version 1.1;
    }
}

4. Configure Accounts & API Keys

• Platform: openai

• Name: Your preferred name.

• Self-Usage Rate: Set to 1.

1. Go to "Account Management" -> "Add Account."

2. Select OpenAI platform and OAuth authorization method.

3. Crucial Step: Clear the default model list. This allows the proxy to automatically recognize new models (like GPT-5.5) without manual updates.

4. Set concurrency (e.g., 10-100 depending on your account's capability).

5. Generate the authorization link and open it in an Incognito browser window.

6. Log in with your ChatGPT credentials (or Team workspace).

7. Note: After login, the page may show an error or go blank. This is normal. Copy the entire URL from the browser's address bar and paste it back into the sub2api confirmation box.

5. Testing & Integration

• cc-switch: A powerful routing tool to manage multiple keys.

• Codex CLI, openclaw, Hermes: Configure these tools to use your self-hosted proxy endpoint.