Connecting in Claude Code

1. Open or create the file ~/.claude/settings.json, where ~ is your home directory

2. Fill it with the contents below

   ~/.claude/settings.json

   {
     "env": {
       "ANTHROPIC_BASE_URL": "<YOUR_URL>",
       "ANTHROPIC_API_KEY": "<YOUR_API_KEY>",
       "CLAUDE_CODE_ENABLE_GATEWAY_MODEL_DISCOVERY": 1
     }
   }

   • Replace <YOUR_API_KEY> with the API key issued to you
     • If you see these instructions but do not have an API key, contact your organization's administrator
   • Replace <YOUR_URL> with the OpenAI-compatible URL issued to you
     • If you see these instructions but do not have an OpenAI-compatible URL, contact your organization's administrator
   • Optional: add the line "CLAUDE_CODE_AUTO_COMPACT_WINDOW": X to the env block — it enables automatic compaction when the context fills up to X tokens. Replace X with the value you need

3. Start Claude Code. When asked Do you want to use this API key?, answer Yes

   • If you see the error API Error: 401 Unknown API key: ..., contact your organization's administrator to reissue your API key

4. To choose a model, use /model. Explyt supports only the models marked From gateway

   • auto – maximum availability (uptime) and high quality
   • Anthropic/opus-5 – Anthropic model with high-quality answers
   • OpenAI/gpt-5.6-sol – general-purpose OpenAI model

If you run into problems while connecting, try typing /logout in the chat and start over.

For errors, see Error messages.