OpenClaw Setup

Difficulty ⌨️ Advanced　|　API format OpenAI-compatible　|　base_url https://byesu.com/v1

Setup · fill these in

API addresshttps://byesu.com/v1

API keysk-xxxxxxxxCreate in console →

Modelgpt-5.5

API format OpenAI-compatible · model names follow the console

OpenClaw (🦞) is an open-source personal AI assistant that lets you use a custom model provider. Point it at byesu — an AI API gateway with OpenAI-compatible and Anthropic-native endpoints — and one key gives you Claude, GPT-5.6, Grok, Gemini and more, pay-as-you-go with no subscription. It works the same on Windows, macOS, and Linux.

Before you start

First go to Console → Tokens and create a token (see Choosing a group).

Option 1: The wizard (easiest)

After installing, run the built-in wizard and choose the "Custom / OpenAI-compatible" provider when prompted:

openclaw onboard

Fill in, one by one:

• API endpoint / baseUrl: https://byesu.com/v1
• API key: your sk-xxxx
• Model: enter a model name available on byesu (such as gpt-5.5 or claude-opus-4-8)

Option 2: Edit the config file by hand

In OpenClaw's config file (~/.openclaw/openclaw.json, in JSON5 format), add a custom provider under models.providers:

{
  agents: {
    defaults: { model: { primary: "byesu/gpt-5.5" } },   // which model to use by default
  },
  models: {
    mode: "merge",                   // merge with the built-in providers instead of replacing them
    providers: {
      byesu: {
        baseUrl: "https://byesu.com/v1",
        apiKey: "${BYESU_API_KEY}",      // use an environment variable; avoid plain text
        api: "openai-completions",       // OpenAI-compatible
        models: [
          { id: "gpt-5.5", name: "Byesu GPT-5.5" },
          { id: "claude-opus-4-8", name: "Byesu Claude Opus 4.8" },
        ],
      },
    },
  },
}

A few key points

• The baseUrl ends in /v1 (OpenAI-compatible).
• For apiKey, use the environment variable ${BYESU_API_KEY} (put the real key in an environment variable) — don't commit it to git in plain text: export BYESU_API_KEY="sk-xxxx".
• If the Gateway runs as a background service / daemon, make sure the variable is also present in the service process's environment (an export in your current terminal alone won't be visible to it, and you'll get a 401). If it's local-only and never goes into git, you can also put the key in the config in plain text.
• The id values under models must match the model names in the Console.
• The exact config file path is whatever the openclaw onboard wizard generates (some versions use ~/.config/openclaw/).

Verify the setup

openclaw models list     # confirm the byesu/-prefixed models appear in the list

You can also run openclaw dashboard to open the UI and send a quick test message.

Stuck?

When an error makes no sense, first run the gateway in the foreground with openclaw gateway --port 18789 and watch the logs and errors directly.

• 401 / invalid token → wrong key, or a background service can't read the BYESU_API_KEY environment variable (see the key points above) (details)
• No available channel → the model isn't in your token's group (Choosing a group, details)
• For anything else, see Common errors & fixes