Apple Turns Siri Into a Physical 3D Orb in visionOS 27

Siri is now a physical object in your room. In visionOS 27, Apple has turned its voice assistant into a persistent 3D orb that you can grab in your hand, pin to a physical desk, and wake up simply by making eye contact. No wake words, no floating menus. It is the defining feature of a software update that redesigns the Vision Pro around Apple's new foundation AI models, rolling out globally today.

The Physics of 180 Seconds

Standard spatial computing relies on foveated rendering—blurring your peripheral vision while sharpening exactly where your eyes focus—to save battery and processing power. visionOS 27 introduces a high-quality spatial recording mode that discards this trick, forcing the headset to render the entire field of view in full 4K resolution edge-to-edge.

The computational cost of running the equivalent of multiple high-end monitors without visual shortcuts is so immense that Apple strictly capped the feature at 180 seconds. It is a precise measurement of exactly where mobile processing power sits today against the realities of hardware physics. But that processing ceiling is also exposing the first real divide between Vision Pro generations.

The Silicon Divide and the EU Exception

The people who spent $3,500 on the original M2 Vision Pro in 2024 are receiving the bulk of this update, but they are also hitting the wall of AI hardware cycles. The M2 chip cannot run Apple's most advanced on-device model, AFM 3 Core Advanced. As a result, early adopters will miss out on the newest voice customization features and rely more heavily on Private Cloud Compute, trading on-device speed for slight latency.

“One nice benefit of Siri AI in visionOS is that it's rolling out in the EU too. So EU users who won't get the new Siri yet on iPhone... can still experience it on Vision Pro.”— Ryan Christoffel

Because the Vision Pro operates under different platform scale metrics than the iPhone, Apple bypassed its current regulatory disputes in the European Union. EU headset owners get full access to the AI tools on day one. With English-only support at launch, the question becomes what these users can actually do with a localized AI model in their physical workspace.

Collapsing the 3D Workflow

The friction of moving 3D work from a computer into the physical world just collapsed. With a new Spatial Preview framework, an industrial designer can click a 3D file in their Mac's Finder and instantly spawn it as a full-scale object inside their physical office. They no longer need to build or compile a native app just to view a prototype.

Apple is applying the same spatial translation to everyday photography. A new feature uses on-device AI to add depth and dimension to standard 2D panoramas, stretching them into workspaces that users can physically step inside. The headset is no longer just a screen you wear; it is a room that understands exactly what you put inside it.

What people are saying

pdfu@itspdfu

“iOS 27 and macOS Golden Gate have private hooks that let apps add Siri Extensions and replace Siri AI’s server backend with third-party models. First, here’s Claude running as a Siri Extension using the Model Delegation API in App Intents. Claude can appear in Siri’s “Ask…””

3.1K2722.4K1.0MSep 13

Mark Gurman@markgurman

“The feature is called Recaps. It lives in the Siri app. It can run all day long, on demand or if you’re at certain places. I believe it will require new hardware that has an updated chip to secure the audio processing. It’s like the Amazon Bee.”

1.4K82156315.2KSep 9

Konrad Muzyka - Rochan Consulting@konrad_muzyka

“The most important effect of the Ukrainian counterattack north of Lyman is that it thwarted Russia’s attempt to envelop the Slovyansk–Kramatorsk agglomeration from the north. We assess that Russian forces have lost control of approximately 240 km² there. By comparison, they have”

1.1K154107198.3KSep 10