How to Change AI Model in OpenClaw (Claude, GPT-4, Gemini)

Quick switch via chat: Type /model in any OpenClaw chat (WhatsApp, Telegram, Discord, or CLI). You'll see a list of configured models. Select the one you want. This changes the model for your current session only.

Switch with command: Send /model claude-opus-4-6 or /model gpt-4 to switch directly. Use /model default to reset to your configured default.

Per-session model: Each conversation can use a different model. Your Telegram chat can use Claude while your WhatsApp uses GPT-4.

Change default model: Edit ~/.openclaw/openclaw.json and set agent.model to your preferred model ID. Example: { "agent": { "model": "anthropic/claude-sonnet-4" } }. Restart the gateway for changes to take effect.

Add new providers: To use models from different providers, add API keys in your config under models.providers. OpenClaw supports Anthropic, OpenAI, Google, MiniMax, OpenRouter, and more.

Model aliases: Set up short aliases like opus, sonnet, gpt4 for quick switching. Configure in models.aliases in your config.

Cost optimization: Use expensive models (Opus) for complex tasks and cheaper models (MiniMax, Sonnet) for simple queries. The /model command makes switching instant.

Best models by task: Claude Opus for complex reasoning, Claude Sonnet for coding, GPT-4 for creative writing, Gemini for speed, MiniMax M2.1 for cost-effective general use.

Easiest option: Use OpenClaw Cloud at cloud.getopenclaw.ai — all models pre-configured, switch with one click, no API keys to manage.