Skip to main content
The Quantum Dispatch
Back to Home
Cover illustration for iOS 27 Siri: What the New Assistant Can Actually Do

iOS 27 Siri: What the New Assistant Can Actually Do

Apple shipped iOS 27 on September 14 with a rebuilt Siri, a new System Orchestrator and 250+ changes. Here is what the assistant actually does now.

Dr. Nova Chen
Dr. Nova ChenSep 15, 20265 min read

Apple Ships the Siri Rebuild It Has Been Promising

Apple released iOS 27 and macOS Golden Gate 27 on September 14, 2026, and the headline feature is the redesigned Siri that Apple has been describing for the better part of two years. This is not a tuning pass. Apple says the assistant runs on new models and is wired together by a component it calls the System Orchestrator, which routes a request across Apple Intelligence features rather than treating each one as a separate destination. Ars Technica and CNBC both covered the rollout on launch day, and AppleInsider counted more than 250 changes across the release.

  • Release date: iOS 27, iPadOS 27 and macOS Golden Gate 27 shipped September 14, 2026
  • New in Siri: typed input, conversation history, follow-up questions, and access to personal context from apps like Messages, Mail and Journal
  • Visual Intelligence: now reachable through a dedicated Siri mode inside the Camera app, and screenshots can be attached as context
  • Requirements: Apple Intelligence needs an iPhone 15 Pro or later; iOS 27 itself installs on iPhone 11 and later and iPhone SE 2nd generation and later

What Can the New Siri Do That the Old One Couldn't?

Three things stand out. The first is that Siri finally remembers. Conversation history means a follow-up question no longer starts from a blank slate, so "what about Thursday?" resolves against what you just asked rather than being parsed as a fresh, contextless request. The second is typed input, which sounds small and is not — it makes the assistant usable in a quiet room, on a train, or anywhere speaking to a phone feels awkward.

The third is personal context. Apple says Siri can now reference information held in Messages, Mail, Journal and other supported apps when it answers. That is the capability that turns an assistant from a search front-end into something that can answer questions about your own life, and it is also the capability with the highest bar for getting right. Apple's stated approach here remains on-device processing wherever the model fits, with Private Cloud Compute handling the rest.

Visual Intelligence Moves Into the Camera

Visual Intelligence is no longer a separate mode you have to go looking for. In iOS 27 it lives as a dedicated Siri mode inside Camera, and you can hand Siri a screenshot or a captured image as context for a question. Point at a poster and ask about the date; screenshot a receipt and ask what you spent. The interaction model is the interesting part — the assistant is being positioned as something you show things to, not just something you talk to.

This follows the direction Apple set out earlier in the year when it opened Apple Intelligence to third-party models. Our earlier coverage of Apple Intelligence adding Claude, ChatGPT and Gemini as options explains how that choice layer fits alongside Apple's own models, and our piece on the ground-up Siri rebuild covers the architecture work that led here.

Which iPhones Get Apple Intelligence in iOS 27?

The split matters. iOS 27 installs on the same range of devices as iOS 26 — iPhone 11 and later, plus iPhone SE 2nd generation and later — but Apple Intelligence, including the new Siri, requires an iPhone 15 Pro or Pro Max, or an iPhone 16 or later. Owners of older supported phones get the rest of the release, which is substantial on its own: Photos gains an Extend tool that generates content past the original frame and a Spatial Reframe feature for shifting perspective, Image Playground adds photorealistic styles, Liquid Glass gets readability work and sharper icons, and HomeKit cameras pick up 4K streaming and recording.

Why an On-Device Assistant Release Matters

The broader story is about where AI inference happens. A cloud assistant is easy to build and expensive to run at a billion-device scale; an on-device assistant is hard to build and nearly free to run. Apple has spent two release cycles pushing work down onto the Neural Engine, and the System Orchestrator is the piece that lets a small on-device model hand off only what it cannot handle. That pattern — a local router in front of larger models — is showing up across the industry, from desktop assistants to Gemini's Windows app. Apple is now running it at the largest install base anyone has tried it on. More coverage of on-device models is in our artificial intelligence section.

Sources: Ars Technica — September 14, 2026; CNBC — September 14, 2026; AppleInsider — September 9, 2026.

More AI Stories