What AI models does Simple Chat use?

OpenAI GPT models, routed by task. You don't manage API keys or pick a model for every customer message; Simple Chat sends each piece of work to the model that fits it.

Why more than one model

Customer-facing replies run on capable models, because accuracy on product, policy, and order questions is worth the extra depth when replies are generated from your store context. Supporting tasks like translation and context selection run on faster settings, where speed matters more than reasoning power. The messages customers actually read get the depth; the plumbing gets the speed.

GPT-5.6 Sol on the Premium plan

Stores that want a higher-capability model can switch to the Premium plan in Billing. On Premium, primary customer chat and supporting tasks currently run on OpenAI GPT-5.6 Sol, and each AI resolution bills at $1.89 instead of $0.29. Faster greeting and extra-fast replies stay on the standard model so those turns stay responsive. If you switch plans, Shopify will ask you to approve the new per-resolution rate before it takes effect.

When a model request fails

Provider outages happen. Simple Chat retries automatically, so a brief hiccup doesn't leave a conversation stuck. If a request still can't complete after the retries, the customer sees a clear recovery message rather than a chat that hangs indefinitely.

What this means for you

Nothing to operate. Model selection, request routing, and keeping pace with provider changes all happen inside Simple Chat, so you get current model capability without running any model infrastructure. Your team's attention goes to instructions, tools, and conversation quality instead of deciding which API to call for each message type.