Recent code analysis of iOS 27 and macOS Golden Gate private frameworks reveals that Apple has engineered a highly extensible architecture for Siri, facilitating deep integration with third-party large language models (LLMs). Through a mechanism called "Model Delegation," developers can implement extensions—such as the existing ChatGPT integration—that allow models like Claude to function as Siri extensions. This enables a bidirectional workflow where a third-party model interprets natural language and then hands tasks back to Siri to execute system-level actions, such as creating Reminders or managing CSV files. More significantly, the "Model Manager Services" framework contains an "Inference Providing" protocol that could allow third-party models, such as GPT-5.6, to entirely replace Apple’s native server-side Siri model. In this advanced configuration, the external LLM receives Apple's specific Siri planner prompts and tool definitions, allowing it to perform complex multi-step sequences like summarizing emails and sending Messages. While the "Ask..." contextual menu is currently limited to ChatGPT in the macOS 27 Release Candidate, the underlying infrastructure suggests a significant shift toward model interoperability, likely influenced by the European Union's Digital Markets Act requirements regarding third-party access to hardware and software features.