Integrate Claude Code

Configure Claude Code through the Nusabase Anthropic Messages-compatible endpoint.

Use this guide to route Claude Code model calls through Nusabase. Claude Code appends /v1/messages to ANTHROPIC_BASE_URL, so configure the root URL: https://api.nusabase.io.

When To Use This

• You primarily use Claude Code from the terminal.
• You want to keep the Anthropic Messages request format.
• You want API keys, quota, and usage to go through Nusabase.

Prerequisites

• Claude Code is installed.
• You have a Nusabase API key: sk-....
• Your machine can reach https://api.nusabase.io.
• The example model qwen3.5-flash appears in the current pricing docs. Use /v1/models as the source of truth.

Configure Environment Variables

Linux / macOS / WSL:

export ANTHROPIC_BASE_URL=https://api.nusabase.io
export ANTHROPIC_AUTH_TOKEN=sk-your_api_key
export ANTHROPIC_MODEL=qwen3.5-flash

Windows PowerShell:

$env:ANTHROPIC_BASE_URL="https://api.nusabase.io"
$env:ANTHROPIC_AUTH_TOKEN="sk-your_api_key"
$env:ANTHROPIC_MODEL="qwen3.5-flash"

Configuration notes:

Setting	Correct Value	Notes
ANTHROPIC_BASE_URL	https://api.nusabase.io	Use the root URL, without /v1.
ANTHROPIC_AUTH_TOKEN	sk-...	Claude Code sends this as Authorization Bearer.
ANTHROPIC_API_KEY	sk-...	Optional alternative; used as x-api-key when AUTH_TOKEN is not set.
ANTHROPIC_MODEL	qwen3.5-flash	Use a real model returned by /v1/models.

Persistent Configuration

Write user-level config to ~/.claude/settings.json:

{
  "env": {
    "ANTHROPIC_BASE_URL": "https://api.nusabase.io",
    "ANTHROPIC_AUTH_TOKEN": "sk-your_api_key",
    "ANTHROPIC_MODEL": "qwen3.5-flash"
  }
}

For shared project config, do not include a real API key:

{
  "env": {
    "ANTHROPIC_BASE_URL": "https://api.nusabase.io",
    "ANTHROPIC_MODEL": "qwen3.5-flash"
  }
}

Verify Nusabase

curl https://api.nusabase.io/v1/models \
  -H "Authorization: Bearer sk-your_api_key"

curl https://api.nusabase.io/v1/messages \
  -H "x-api-key: sk-your_api_key" \
  -H "anthropic-version: 2023-06-01" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3.5-flash",
    "max_tokens": 256,
    "messages": [{"role": "user", "content": "hello"}]
  }'

Start Claude Code

claude

In the interactive session, send:

Please summarize the current project structure.

Model Selection Notes

Claude Code gateway model discovery only adds some Anthropic-style model IDs to the model picker. For non-Claude names such as qwen3.5-flash, set ANTHROPIC_MODEL explicitly.

To try gateway model discovery:

export CLAUDE_CODE_ENABLE_GATEWAY_MODEL_DISCOVERY=1

Troubleshooting

Problem	Likely Cause	Fix
404	ANTHROPIC_BASE_URL includes /v1	Use https://api.nusabase.io.
401	API key is wrong or sent with the wrong header	Prefer ANTHROPIC_AUTH_TOKEN=sk-..., then verify with curl.
Model unavailable	ANTHROPIC_MODEL is not enabled for your account	Call /v1/models, then copy an available model name.
qwen model not shown in picker	Claude Code gateway discovery filters model IDs	Set ANTHROPIC_MODEL explicitly.

Next, see First Request Example.

References

• Claude Code LLM gateway
• Claude Code overview
• Anthropic Messages API
• First Request Example
• Error Codes