Claude Code Setup in China: Complete Guide — from Registration to First Autocompletion in 10 Minutes

Claude Code is widely recognized as one of the best AI coding tools available, but in Hong Kong and China the first step is often where it stops: official sign-up does not support your region. Fortunately, Claude Code itself supports a custom API address. This guide walks the whole path — no VPN, no code changes, about ten minutes from start to finish.

Why This Works

Claude Code officially supports pointing its API address elsewhere through environment variables: ANTHROPIC_BASE_URL sets the server and ANTHROPIC_AUTH_TOKEN supplies the credential. Point both at Essevin and Claude Code runs on your Essevin account. Essevin is directly accessible from mainland China without a proxy.

It is still the official CLI. The only things you change are the address it connects to and the key it uses.

Three-Step Setup

First, register a free account and sign in at console.essevin.com. Essevin runs on a monthly subscription and subscriptions are now open; choose a plan on the Subscription page of the console, then follow the steps below.

Second, open API keys, click New key, enter a name (say, claude-code), pick Claude under Use with, and create the key. It begins with sk-.

Third, configure Claude Code in one of two ways:

• One-click setup (recommended): select One-Click Setup beside the key, choose Claude Code and your system (macOS / Linux or Windows PowerShell), copy the command and paste it into your terminal. The script installs Claude Code if needed, backs up and writes ~/.claude/settings.json, and verifies the key; a green success message means you are done. The command carries a single-use integration token that expires after 15 minutes, not the API key itself.
• Manual setup: set two environment variables in your terminal:

# Point Claude Code at Essevin (no /v1 in the address)
export ANTHROPIC_BASE_URL=https://api.essevin.com
export ANTHROPIC_AUTH_TOKEN=sk-your-key

claude   # Use normally; usage shows up in the console usage records

Run claude and start a conversation. If it replies normally, you are connected. To make a manual setup permanent, add the two lines to ~/.zshrc or ~/.bashrc; on Windows, use setx and reopen the terminal. If you prefer a graphical tool or switch between several providers, use CC Switch; the full reference is in the Claude Code setup docs.

Three Common Pitfalls

• Environment variables not taking effect: After editing your shell config, open a new terminal or run source; terminals launched from your IDE might not read your shell config.
• An extra /v1, or the wrong key: Claude Code uses the Anthropic protocol, so the address is https://api.essevin.com without /v1 (otherwise requests go to /v1/v1/messages). The key must be one created for Claude; a GPT key does not work with Claude Code. Both mistakes return a 404.
• Corporate proxy interference: Enterprise HTTPS proxies sometimes rewrite request headers. If you see weird 401/403 errors, test in an environment without a proxy first.

Switch Models by Task

A Claude key can use the Claude models included in your plan. Type /model inside a session, or start with --model:

These IDs are examples; the list returned by GET https://api.essevin.com/v1/models for your key is what counts.

Verification and Transparent Usage

After one request, the console usage records show it immediately — model, token count and the points deducted, all aligned. Essevin is a flat monthly subscription: chat, images and video share one monthly points pool, deducted by actual usage and reset every month, with no pay-as-you-go balance. Tokens that hit prompt cache are counted at the lower cache rate, and conversational tools like Claude Code usually see a high share of cache hits, so the points you actually use often come in below what list prices suggest. Subscriptions are now open; compare the plans at essevin.com, then register for free and subscribe at console.essevin.com.