Skip to main content

Claude Code CLI

Connect Claude Code very quickly to your instance of Dedicated Gateway and use IG1 AI soveraing model with Claude Code.

Environment variables​

The easiest way to connect Claude Code to IG1 AI Dedicated gateway is to use Environment variables:

Set following variables

  • ANTHROPIC_AUTH_TOKEN with your API KEY,
  • ANTHROPIC_BASE_URL with your instance (e.g. https://api.xxx.ig1.ai),
  • ANTHROPIC_MODEL set your prefered LLM or Virtual Model
  • CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC (optional) - Disable Claude code traffic for metrics collection.

Example:

export ANTHROPIC_AUTH_TOKEN='sk-XXXX'
export ANTHROPIC_BASE_URL='https://api.xxx.ig1.ai'
export ANTHROPIC_MODEL='coding-max-thinking'
export CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC=1
note

coding-max-thinking is a virtual model available on our demo environment. Its routes to glm53-flash-agent. Feel free to use other coding models from this list.

Open Claude Code​

After setting the environment variable, launch Claude Code to test if it sees the new default model

$ claude
╭─── Claude Code v2.1.119 ────────────────────────────────────────────────────────────────────────────╮
│                                 │ Tips for getting started                                          │
│          Welcome back!          │ Run /init to create a CLAUDE.md file with instructions for Claude │
│                                 │ ───────────────────────────────────────────────────────────────── │
│             ▐▛███▜▌             │ Recent activity                                                   │
│            ▝▜█████▛▘            │ No recent activity                                                │
│              ▘▘ ▝▝              │                                                                   │
│                                 │                                                                   │
│   IG1 GPT · API Usage Billing   │                                                                   │
│     ~/Documents/claude-test     │                                                                   │
╰─────────────────────────────────────────────────────────────────────────────────────────────────────╯

❯ /model

──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
  Select model
  Switch between Claude models. Applies to this session and future Claude Code sessions. For other/previous model names, specify with --model.

    1. Default (recommended)  Use the default model (currently Sonnet 4.6) · $3/$15 per Mtok
    2. Sonnet (1M context)    Sonnet 4.6 for long sessions · $3/$15 per Mtok
    3. Opus 4.7 (1M context)  Opus 4.7 with 1M context · Most capable for complex work · $5/$25 per Mtok
    4. Haiku                  Haiku 4.5 · Fastest for quick answers · $1/$5 per Mtok
  ❯ 5. IG1 GPT ✔              Custom model

  ● High effort (default) ← → to adjust

  Use /fast to turn on Fast mode (Opus 4.6 only).

  Enter to confirm · Esc to exit

🎉 It works! Claude uses coding-max-thinkingas default model.