Claude Code CLI

Configure Anthropic's Claude Code CLI to route through LLMAI for unified billing and model access.

Model slugs in examples are illustrative. Confirm current availability in the live catalog and your account's /v1/models before use.

What This Does

Claude Code is Anthropic's official CLI for AI-assisted development. By configuring it to use LLMAI's endpoint, you can use any supported model — not just Claude — while consolidating all your AI spend under a single LLMAI account. Your token usage draws from your LLMAI balance automatically.

Setup

Step 1: Locate Your Settings File

Claude Code stores its configuration in a settings.json file. Find it based on your OS:

OS	Path
macOS / Linux	`~/.claude/settings.json`
Windows	`%USERPROFILE%\.claude\settings.json`

If the file doesn't exist, create it.

Step 2: Configure Environment Variables

Open settings.json and add the following configuration:

{
  "env": {
    "ANTHROPIC_BASE_URL": "https://api.llmai.dev/",
    "ANTHROPIC_AUTH_TOKEN": "YOUR_LLMAI_API_KEY",
    "ANTHROPIC_MODEL": "gpt-5.6-sol",
    "ANTHROPIC_DEFAULT_SONNET_MODEL": "kimi-k2.7-code",
    "ANTHROPIC_SMALL_FAST_MODEL": "glm-5-turbo"
  }
}

Replace YOUR_LLMAI_API_KEY with your actual key from console.llmai.dev/keys.

Step 3: Alternative — Environment Variables in Shell

If you prefer not to modify the settings file, set environment variables in your shell:

export ANTHROPIC_BASE_URL="https://api.llmai.dev/"
export ANTHROPIC_AUTH_TOKEN="YOUR_LLMAI_API_KEY"

Add these to your ~/.zshrc or ~/.bashrc to persist across sessions.

Step 4: Verify the Connection

Run Claude Code and start a conversation:

claude

Try a simple prompt to confirm the connection works.

Recommended Models for Claude Code

Use Case	Model	Why
Complex reasoning & planning	`gpt-5.6-sol`	OpenAI's GPT-5.6 flagship for complex work
Code generation & refactoring	`kimi-k2.7-code`	Optimized for code
Large codebase analysis	`gemini-3.1-pro`	1M+ token context
Quick operations	`glm-5-turbo`	Fast, low latency
Budget workloads	`glm-5.3-flash`	Cost-efficient coding and agent tasks

Troubleshooting

Connection refused or timeout

Verify the base URL is exactly https://api.llmai.dev/ (with trailing slash for Claude Code)
Check that your API key is valid at console.llmai.dev/keys

Authentication errors

Ensure ANTHROPIC_AUTH_TOKEN is set, not ANTHROPIC_API_KEY — Claude Code expects the token variant
Remove any quotes around the key if you're setting it via shell export

Model not found errors

Use model slugs exactly as shown in the Models reference
Claude Code may require you to specify a model explicitly if the default isn't recognized

Notes

Claude Code was designed for Claude models but works with any OpenAI-compatible endpoint
The ANTHROPIC_SMALL_FAST_MODEL is used for background tasks — set it to a low-cost model
All token consumption appears in your LLMAI dashboard in real-time

On this page