Claude / Claude Code Configuration

Claude Code is Anthropic's official terminal command-line AI coding assistant. Connect it through the 4AICode API to manage keys, team quotas, and usage in one console, with billing based on actual token usage.

💡 Beginner tip: We recommend starting with [Method 1: Quick Script Installation & Setup]. A single command handles everything automatically, saving time and effort.

About Route Selection: 4AICode provides two interchangeable network routes with shared accounts and tokens:

• International Route (Recommended for global users): Uses https://relay.4aicode.com.
• Mainland China Direct Route: Uses https://relay.4aicode.net (more stable for direct connections in mainland China).

Prerequisites: Get Your Token

Whichever setup method you choose, you first need your 4AICode token:

1. Sign in to 4AICode and go to Dashboard → Manage Tokens (or 4aicode.net/tokens in mainland China).
2. Click "New Token".
3. Copy the generated string (formatted like sk-4ai-<your-token>).

⚠️ Important security notes:

• The token is shown only once upon creation. Save it to a password manager or notes app.
• All sk-4ai-<your-token> strings in this guide are placeholders. Replace them with your actual token.
• Treat your token like a password: never share it or commit it to GitHub. Tokens typed directly into shell commands remain in terminal history; clear your history if necessary (e.g. by running history -c or clearing the terminal history file).

Method 1: Quick Script Installation & Setup (Recommended for Beginners)

Select the tab matching your operating system:

# Official installer command
curl -fsSL https://claude.ai/install.sh | bash

# International route (recommended for overseas/global network)
cat <<'EOF' >> ~/.zshrc
export ANTHROPIC_BASE_URL=https://relay.4aicode.com
export ANTHROPIC_AUTH_TOKEN=sk-4ai-<your-token>
export ANTHROPIC_API_KEY=sk-4ai-<your-token>
EOF
source ~/.zshrc

# Note: If you use bash instead of zsh (e.g., on some Linux distros), replace ~/.zshrc with ~/.bashrc

claude

Method 2: Global Configuration File (Cleaner Approach)

In addition to environment variables, Claude Code supports global configuration via settings.json. This approach keeps your shell environment clean and persists reliably.

Configuration file path:

• macOS / Linux: ~/.claude/settings.json
• Windows: %USERPROFILE%\.claude\settings.json

Open or create this file, and insert the configuration under the env key (if existing settings are present, preserve them and merge the env object):

{
  "env": {
    "ANTHROPIC_BASE_URL": "https://relay.4aicode.com",
    "ANTHROPIC_AUTH_TOKEN": "sk-4ai-<your-token>",
    "ANTHROPIC_API_KEY": "sk-4ai-<your-token>"
  }
}

Save and restart your terminal or Claude Code for the changes to take effect.

Advanced: Using Other Top-Tier Models in Claude Code

The 4AICode backend gateway provides intelligent model routing. You can use Claude Code's polished terminal interface while running ChatGPT (GPT-5.6 / GPT-6) or Kimi K3 under the hood!

0. Using Claude Opus 5.5 Kiro in Claude Code (Best Value)

Claude Opus 5.5 Kiro is 4AICode's exclusive Claude Code-optimized version, offering exceptional value with discounted pricing.

# Using Claude Opus 5.5 Kiro (Recommended)
export ANTHROPIC_BASE_URL=https://relay.4aicode.com
export ANTHROPIC_AUTH_TOKEN=sk-4ai-<your-token>
export ANTHROPIC_DEFAULT_OPUS_MODEL=claude-opus-5-5-kiro
export ANTHROPIC_MODEL=opus
claude

Use Cases:

• Complex code generation and architecture design
• Multi-step reasoning and problem solving
• Large-scale context handling (200K tokens)
• Technical documentation writing and code review

1. Using ChatGPT Models in Claude Code

Simply set the model environment variables. When receiving /v1/messages requests, the gateway selects the upstream service corresponding to the requested gpt-* model:

# Recommended daily driver: GPT-5.6 Terra
export ANTHROPIC_BASE_URL=https://relay.4aicode.com
export ANTHROPIC_AUTH_TOKEN=sk-4ai-<your-token>
export ANTHROPIC_DEFAULT_OPUS_MODEL=gpt-5.6-terra
export ANTHROPIC_MODEL=opus
claude

# Top-tier flagship: GPT-6 Astra
export ANTHROPIC_DEFAULT_OPUS_MODEL=gpt-6-astra
export ANTHROPIC_MODEL=opus
claude

# Complex architecture & deep reasoning: GPT-5.6 Sol with ultra-high reasoning
export ANTHROPIC_BASE_URL=https://relay.4aicode.com
export ANTHROPIC_AUTH_TOKEN=sk-4ai-<your-token>
export ANTHROPIC_DEFAULT_OPUS_MODEL=gpt-5.6-sol
export ANTHROPIC_DEFAULT_OPUS_MODEL_NAME="4AICode GPT 5.6 Sol"
export ANTHROPIC_DEFAULT_OPUS_MODEL_SUPPORTED_CAPABILITIES=effort,xhigh_effort,thinking,adaptive_thinking,interleaved_thinking
export ANTHROPIC_MODEL=opus
export ANTHROPIC_DEFAULT_HAIKU_MODEL=gpt-5.4-mini
export CLAUDE_CODE_EFFORT_LEVEL=xhigh
claude

2. Using Kimi K3 in Claude Code (Ultra-Long 1M Context)

If you need to analyze massive codebases or large documentation sets, connect Kimi K3. See the dedicated guide: 👉 Detailed Guide: Using Kimi K3 with Claude Code

Other Tools & SDK Integration

4AICode is fully compatible with official Anthropic API specifications, enabling seamless integration with any Anthropic-supported SDK or tool:

1. Unified Connection Parameters

• Base URL: https://relay.4aicode.com (use https://relay.4aicode.net in mainland China)
• API Key / Token: Your 4AICode Token (sk-4ai-<your-token>)

2. Node.js / TypeScript SDK

import Anthropic from "@anthropic-ai/sdk";

const client = new Anthropic({
  baseURL: "https://relay.4aicode.com",
  apiKey: "sk-4ai-<your-token>",
});

const msg = await client.messages.create({
  model: "claude-sonnet-4-6",
  max_tokens: 1024,
  messages: [{ role: "user", content: "Hello!" }],
});

console.log(msg.content);

3. Python SDK

import anthropic

client = anthropic.Anthropic(
    base_url="https://relay.4aicode.com",
    api_key="sk-4ai-<your-token>",
)

message = client.messages.create(
    model="claude-sonnet-4-6",
    max_tokens=1024,
    messages=[
        {"role": "user", "content": "Hello, please write a quicksort algorithm."}
    ]
)

print(message.content[0].text)

4. cURL Request Example

4AICode supports both the Anthropic official x-api-key header and the standard Authorization: Bearer header:

# Option A: Using x-api-key (Anthropic standard)
curl https://relay.4aicode.com/v1/messages \
  -H "x-api-key: sk-4ai-<your-token>" \
  -H "anthropic-version: 2023-06-01" \
  -H "content-type: application/json" \
  -d '{
    "model": "claude-sonnet-4-6",
    "max_tokens": 128,
    "messages": [{"role": "user", "content": "ping"}]
  }'

# Option B: Using Authorization: Bearer
curl https://relay.4aicode.com/v1/messages \
  -H "Authorization: Bearer sk-4ai-<your-token>" \
  -H "anthropic-version: 2023-06-01" \
  -H "content-type: application/json" \
  -d '{
    "model": "claude-sonnet-4-6",
    "max_tokens": 128,
    "messages": [{"role": "user", "content": "ping"}]
  }'

Frequently Asked Questions (FAQ)

Q1: claude: command not found or command unrecognized?

• Cause: Claude Code is not installed, or your terminal PATH environment variable has not updated.
• Fix: Completely close and reopen your terminal. If the issue persists:
  • macOS/Linux: Re-run the official installer curl -fsSL https://claude.ai/install.sh | bash and ensure ~/.local/bin is in your PATH.
  • Windows: Run the setup script in PowerShell: irm https://4aicode.com/claude-install.ps1 | iex, which automatically configures PATH.

Q2: Encountered 401 Unauthorized or invalid token?

• Cause: Token string is incorrect, missing characters during copying, or placeholder was not replaced.
• Fix: Visit Dashboard → Manage Tokens to inspect or recreate your token. Ensure it begins with sk-4ai-.

Q3: Why didn't environment variable or settings.json changes take effect?

• Cause: The current terminal process has cached previous variables, or settings.json takes priority over environment variables.
• Fix: Completely exit the running Claude session and restart the terminal. Note that settings.json values override terminal environment variables.

Q4: Common HTTP Error Codes