ai-core's ChatModel on pi-ai: every provider pi-ai knows, with <provider>/<model id> as the model. Through ProxyChatModel the model is pi:<provider>/<model id>.
ChatModel
<provider>/<model id>
ProxyChatModel
pi:<provider>/<model id>
Protected
Readonly
Emits one response per text delta, then a last one with the whole message (text, tool calls, usage).
ai-core's
ChatModelon pi-ai: every provider pi-ai knows, with<provider>/<model id>as the model. ThroughProxyChatModelthe model ispi:<provider>/<model id>.