Apple Confirms AFM Cloud Is Its Own AI Model, Trained With Gemini Outputs

Apple has provided new details about its AI strategy, clarifying that AFM Cloud, short for Apple Foundation Models Cloud, remains an Apple-developed model despite its relationship with Google's Gemini technology. The clarification comes as developers and industry observers try to understand how Apple is building the next generation of Apple Intelligence and Siri.



The announcement helps separate two important parts of Apple's AI effort: the models running directly on devices and the larger models operating through cloud infrastructure.



AFM Local Models Are Fully Apple's Own







According to Apple, the AFM local models that run on iPhones, iPads, and Macs are entirely Apple-built models. These systems are trained, optimized, and deployed by Apple and form the foundation of many Apple Intelligence features available on-device.



Apple has repeatedly emphasized that on-device processing remains a key part of its AI strategy because it delivers faster responses while protecting user privacy. The company previously revealed that its foundation models are trained using Apple's own frameworks, datasets, fine-tuning methods, and reinforcement learning systems.



AFM Cloud Pro Has Gemini Connections







The more interesting clarification involves AFM Cloud Pro, Apple's larger cloud-based model.



Apple confirmed that AFM Cloud Pro is not simply a Gemini model running under Apple branding. Instead, Gemini appears to serve as a foundation and source of training outputs, while Apple performs significant work on top of it.



The company says it handles its own:




Pre-training processes



Post-training optimization



Reinforcement learning



Fine-tuning



Alignment and safety training



Model customization for Apple Intelligence




This means Apple uses Gemini-generated outputs and technology as part of the development pipeline, but the final model delivered to users is still considered an Apple model rather than a direct Gemini deployment.



Siri and Apple Intelligence







The clarification arrives as Apple expands its AI partnership with Google. Earlier announcements confirmed a multi-year collaboration between the two companies, with future Apple Foundation Models benefiting from Gemini technology.



However, Apple wants developers and users to understand that the company still controls the training process, privacy protections, deployment infrastructure, and user experience. Apple Intelligence continues to run across devices and Apple's cloud systems, with Apple maintaining responsibility for model behavior and privacy standards.



For Apple, ownership of the model stack is about more than branding. The company wants to maintain control over privacy, security, and integration across its ecosystem.



The latest clarification suggests Apple is taking a hybrid approach. Rather than building every foundation model entirely from scratch, it is leveraging Gemini's strengths while applying Apple's own training methods and optimization layers. As Apple Intelligence evolves, AFM Cloud Pro appears set to become a central part of that strategy, combining external AI foundations with Apple's in-house model development expertise.