Previously, the playground distinguished models based solely on their URL. However, with the introduction of inference router, multiple models can now share the same URL endpoint. This update changes the model identification logic to use model names instead, allowing the playground to support multiple models behind the same endpoint.