Apple unveiled a major overhaul of Apple Intelligence, a new architecture built on foundation models co‑developed with Google using Gemini technology. Models run on‑device and via Private Cloud Compute, delivering a huge upgrade with state‑of‑the‑art understanding, reasoning, and multimodal image generation capabilities for developers.

The upgraded models enable realistic image creation, advanced photo editing, and visual question answering. Certain devices will receive a higher‑power variant with speech generation, improved dictation, and stronger natural language understanding, though Apple did not name which. A new system orchestrator sits at the core, coordinating features securely across Apple’s ecosystem and tailoring responses to the active app and user task.

Apple framed the move as a privacy‑first alternative to rivals it accused of racing ahead without safeguards. All processing stays on‑device or in Private Cloud Compute, and user data is used only for the immediate request, inaccessible to Apple or third parties. Independent auditors can verify these claims at any time, underscoring Apple’s commitment to secure, system‑wide AI.

The shift arrives as Google rolls out Gemini Intelligence across Android, intensifying the AI arms race. Apple’s on‑device focus gives it a privacy edge and promises richer photo tools and smarter assistants on upcoming iPhone and Mac models. Developers also gain a unified AI layer, potentially setting a new baseline for privacy‑centric intelligence on mainstream hardware.