Skip to main content

Change the default AI model

The default model controls response quality, speed, and cost. Clawify ships with a curated list of models from multiple providers and lets you switch at any time.

Where to find itDirect link to Where to find it​

  1. Open Clawify in your Shopify admin.
  2. Click Assistant settings on the home page (or go to /app/assistant).
  3. Open the AI Model tab.

You will see two sections:

  • Default AI Model — the curated list of models you can pick from right now.
  • Bring Your Own Key (BYOK) — connect your own OpenAI key, xAI key, or ChatGPT subscription.

Pick a built-in modelDirect link to Pick a built-in model​

The built-in list shows every model your current plan can use. Pick the one you want and click Use this model. Your AI agent restarts automatically (~10 seconds) and the new model is applied to every new conversation.

info

Some advanced models (e.g. GPT-5 family, Claude Opus, Grok 4) are gated behind the Growth and Business plans. If a model is locked, the card shows the required plan.

Bring Your Own Key (Growth+ plan)Direct link to Bring Your Own Key (Growth+ plan)​

If you already pay for OpenAI, xAI, or ChatGPT, you can plug those credentials in and skip the AI credits system entirely.

OpenAI / xAI API keyDirect link to OpenAI / xAI API key​

  1. Generate a key on the provider's dashboard:
  2. In Clawify, open Assistant settings → AI Model → Bring Your Own Key.
  3. Click the provider, paste the key, pick which of your available models to expose, then Save.
  4. Your agent restarts. Pick one of the BYOK models from the Default AI Model list above.

Your key is encrypted at rest and used only for direct API calls from your OpenClaw instance.

ChatGPT subscription (Codex OAuth)Direct link to ChatGPT subscription (Codex OAuth)​

If you have a ChatGPT Plus or Pro subscription, you can connect it directly — no API key, no per-token billing. See the dedicated guide: How to use your Codex subscription.

TipsDirect link to Tips​

  • Speed vs. quality trade-off — Smaller models (e.g. gpt-5-mini, claude-haiku-5) reply faster and cost less; bigger ones (gpt-5, claude-opus-5, grok-4) reason better on complex Shopify queries.
  • Per-agent override — You can override the default per sub-agent in Agents if a specialist needs a stronger or cheaper model.
  • Refresh model list — If you just enabled a new model on your provider account, click Refresh in the BYOK row to pick it up without re-entering the key.