Claude Code

Claude Code supports custom gateways through environment variables. Tokener.ai serves the Anthropic Messages API at POST /messages, so Claude Code runs against Tokener.ai with your API key and any chat model from the Models catalog.

Configure

Set the base URL, credential, and model. The base URL is the origin only — no /v1 suffix; Claude Code appends the path itself:

export ANTHROPIC_BASE_URL=https://api.tokener.dev
export ANTHROPIC_AUTH_TOKEN=sk-...   # your Tokener.ai API key
export ANTHROPIC_MODEL=<model-id>

ANTHROPIC_AUTH_TOKEN is sent as the Bearer token; ANTHROPIC_API_KEY (sent as x-api-key) works as well. Create keys in API keys.

To persist the configuration everywhere Claude Code runs, use the env block of ~/.claude/settings.json instead of shell exports:

{
  "env": {
    "ANTHROPIC_BASE_URL": "https://api.tokener.dev",
    "ANTHROPIC_AUTH_TOKEN": "sk-...",
    "ANTHROPIC_MODEL": "<model-id>"
  }
}

Run /status inside Claude Code to confirm the active base URL and credential source.

Background model

Claude Code sends background traffic such as session titles and summaries to a small model. Route it to a model available to your key:

export ANTHROPIC_DEFAULT_HAIKU_MODEL=<model-id>

Expected notices

• Claude Code assumes a 200k context window for model ids it does not recognize and warns once. Set CLAUDE_CODE_MAX_CONTEXT_TOKENS to the model's real context window from the Models catalog.
• Optional: CLAUDE_CODE_ENABLE_GATEWAY_MODEL_DISCOVERY=1 lets the /model picker list the models available to your key.

Claude Code requests bill against your organization's credits like any other traffic on the model API.