Apple Intelligence 2 review: 7 features to try first when iOS 27 ships Sept 14

iOS 27 ships to the public on Monday, September 14, 2026, alongside the iPhone 18 Pro / Pro Max retail launch. The release ships Apple Intelligence 2.0 — Apple's second-generation AI stack, now powered by Google Gemini under a multi-year partnership worth approximately $10 billion per year. The biggest change from Apple Intelligence 1.0 (June 2025) is that Siri 2.0 is conversational, context-aware, and capable of cross-app actions. The biggest limitation is that none of it works on iPhone 15 or earlier.

Here are the seven features worth trying first — and the hardware and regional gating you need to know about before you update.

What changed from Apple Intelligence 1.0?

Three structural changes:

Siri 2.0 is Gemini-backed. The architecture is a three-tier hybrid: on-device 3B-parameter model on the Neural Engine → Apple's Private Cloud Compute servers (no user data retention) → Google Gemini for tasks that exceed on-device capacity. ChatGPT is now opt-in only via "Siri Extensions" — users pick Gemini default, ChatGPT, or Claude.
Live Translation works in 16 languages, with AirPods support for in-person conversations. This was the most-anticipated feature and it ships at launch.
Hardware gating tightened. Apple Intelligence 2.0 requires iPhone 16 or later, iPhone 15 Pro / Pro Max, or any M-series iPad/Mac. iPhone 15 and iPhone 14 owners get iOS 27 (the OS) but not Apple Intelligence 2.0 (the AI).

The Apple-Google Gemini partnership is the biggest single change. Until now, Apple Intelligence was branded as fully proprietary. The reality: Apple's foundation model stack now runs on Google's Gemini, with Apple's on-device and Private Cloud Compute layers on top. Apple-Google deal value: ~$10B/year, comparable to Apple's total App Store revenue from search defaults in some years.

What are the 7 features worth trying first?

1. Live Translation (Messages + FaceTime + Phone + AirPods)

What it does: Real-time translation across Messages (text overlay), FaceTime (live captions one-on-one), Phone (spoken + on-screen text), and AirPods (hands-free in-person translation). All processing is on-device — conversations do not leave the phone.

Languages at launch (16): English, Danish, Dutch, French, German, Italian, Norwegian, Portuguese, Spanish, Swedish, Turkish, Vietnamese, Chinese (Simplified & Traditional), Japanese, Korean. Five more languages (Arabic, Hindi, Polish, Thai, Greek) are scheduled for October-November 2026 updates.

Why it matters: Live Translation is the single best feature in Apple Intelligence 2.0. Tom's Guide called it the standout in a 3-week beta test. It works offline, it works in noisy environments (it uses the AirPods Pro 3's beam-forming mics for in-person mode), and it integrates with Messages — Apple's most-used app globally.

The catch: AirPods Live Translation is not available in the EU at launch under Digital Markets Act constraints. China is pending CAC regulatory approval. Live Translation on the iPhone itself (without AirPods) is available globally but requires iPhone 16 or later.

2. Siri 2.0 — Conversational + Personal Context

What it does: Siri now handles multi-turn conversations (you don't have to repeat the wake word), remembers personal context (it can search WhatsApp, texts, emails), and executes cross-app actions ("summarize the article in Safari, add the todos to Reminders, send the summary to John on WhatsApp").

Why it matters: This is the feature that fixes the decade-long "Siri is bad" complaint. Bip Austin's month-long beta test found the new Siri "dramatically better" — more reliable, less chatty, with a compact bubble UI replacing the glowing-border overlay.

The catch: Personal Context is opt-in. You have to enable "Allow Siri to learn from this app" per app. Most users will leave defaults and miss the feature.

3. Visual Intelligence 2 — Camera Viewfinder AI

What it does: Point your camera at a flyer, poster, restaurant menu, or product — Siri identifies objects, extracts text, looks up products, and adds events to your calendar automatically. New in v2: identify actors from video clips (works with Movies, TV+, Netflix), scan membership barcodes into Wallet passes, and analyze nutritional content from meal photos.

Why it matters: This is the killer AR feature. Tom's Guide called it "the feature that won me over." Apple's pitch: your camera becomes your interface. Point, tap, done.

The catch: Visual Intelligence 2 requires iPhone 15 Pro / Pro Max or later. Older devices get Visual Intelligence 1 with limited object recognition.

4. Reframe for Photos (AI-powered photo editing)

What it does: AI-driven photo reframing that automatically crops, straightens, and recomposes your photos using Apple's Vision Pro spatial technology. It analyzes the subject, the composition rules, and the depth map to suggest three to five alternate crops per photo.

Why it matters: This is the first AI photo feature Apple has shipped that genuinely improves on what you can do manually. Tom's Guide said it won them over specifically because it doesn't oversaturate or over-sharpen — it just finds better compositions.

The catch: Reframe runs on-device but takes 5-20 seconds per photo. If you batch-edit a 1,000-photo library, plan to wait.

5. Playground — Image Generation + Message Backgrounds

What it does: AI image generation in three styles (Animation, Illustration, Sketch) with message-background integration. You can generate an image in Playground and use it as your Messages background, or share it in conversations.

Why it matters: Playground is Apple's answer to DALL-E and Midjourney. The output quality is below Midjourney v7 but above DALL-E 3, per reviewer consensus.

The catch: All generated images include invisible C2PA metadata marking them as AI-generated. Sharing them in Messages preserves the metadata. Screenshotting strips it.

6. Shortcuts AI Automations

What it does: Describe an automation in natural language ("when I leave the office, text my wife I'm heading home, start my commute playlist, and set Do Not Disturb until 6 PM"), and Apple Intelligence generates the full Shortcut. The shortcut is editable — Apple Intelligence creates the steps, you adjust them.

Why it matters: Shortcuts is already one of the most powerful iOS features. AI-generated automations make it accessible to non-technical users.

The catch: Complex automations sometimes need 2-3 prompt iterations to get right. The first generation rarely works perfectly.

7. Live Captions + Voice Memo Transcription

What it does: Live captions in FaceTime (1:1 only at launch, group calls coming October 2026), and AI-powered Voice Memo transcription with speaker labels.

Why it matters: Live Captions in FaceTime is the accessibility win of the release. Voice Memo transcription has been a long-requested feature.

The catch: Voice Memo transcription is on-device but only in English at launch. Spanish, French, German come in October.

What's the hardware gating?

Feature	iPhone 11–14	iPhone 15 / 15 Plus	iPhone 15 Pro / 15 Pro Max	iPhone 16 / 16 Plus	iPhone 16 Pro / 16 Pro Max	iPhone 17 series	iPhone 18 / Air / Pro / Pro Max / Duo
iOS 27 OS	✓	✓	✓	✓	✓	✓	✓
Apple Intelligence 2.0 base	✗	Limited	✓	✓	✓	✓	✓
Live Translation	✗	✗	Some	✓	✓	✓	✓
Visual Intelligence 2	✗	✗	✓	✓	✓	✓	✓
Custom Voice (Siri AI)	✗	✗	✗	✗	✓	✓	✓
On-device 3B model	✗	✗	Limited	✓	✓	✓	✓
A20 Pro NPU	N/A	N/A	N/A	N/A	N/A	N/A	✓ (best perf)

Bottom line: If you have an iPhone 15 or earlier, you get iOS 27 but lose most of Apple Intelligence 2.0. If you have an iPhone 15 Pro / Pro Max, you get base Apple Intelligence 2.0 but lose the best features (Custom Voice, on-device 3B model). To get the full stack, you need iPhone 16 Pro / Pro Max or later — and the iPhone 18 Pro / Pro Max / Duo deliver the best performance because of the A20 Pro's 38 TOPS NPU.

What regions get which features?

Region	iOS 27 OS	Apple Intelligence 2.0	Live Translation (iPhone)	Live Translation (AirPods)
United States	✓	✓	✓	✓
Canada	✓	✓	✓	✓
UK	✓	✓	✓	✓
EU	✓	Limited (DMA)	✓	✗
China	✓	Limited (CAC)	Pending	✗
India	✓	✓	✓	✓
Japan	✓	✓	✓	✓
Korea	✓	✓	✓	✓
Australia	✓	✓	✓	✓

EU limitations (Digital Markets Act): Apple Intelligence 2.0 ships in the EU but with reduced features — sideloading opened April 14, 2026, €500M fine April 15, gatekeeper appeal dismissed July 8. Apple is reportedly working on browser-engine porting compliance but it remains unresolved as of Sept 11, 2026.

China limitations: Apple Intelligence 2.0 uses Baidu ERNIE as the foundation model (not Gemini) per Apple's Baidu partnership. CAC approval is conditional; some features (Visual Intelligence 2, Custom Voice) are not yet approved for Chinese iPhones. Most Live Translation languages are limited to English-Chinese and English-Cantonese at launch.

How does Apple Intelligence 2.0 compare to Google Gemini and ChatGPT?

The honest answer is: it's competitive but not best-in-class.

On-device tasks (Reframe, Voice Memo transcription, Live Captions): Apple wins. The on-device 3B model is faster and more private than cloud-only competitors.
Conversational AI (Siri 2.0): Roughly tied with ChatGPT 5 / Gemini 2.5. Multi-turn dialogue works well. Personal context integration is best-in-class because of Apple's platform access (Messages, Mail, Photos, Calendar).
Image generation (Playground): Below Midjourney v7, roughly tied with DALL-E 3. The output is good but not industry-leading.
Live Translation: Above Google Translate and DeepL for in-person AirPods use. Comparable to dedicated translation apps offline.

The Apple advantage is integration, not raw capability. Apple Intelligence 2.0 is "good enough" at most tasks and "best in class" at the things only Apple can do (cross-app actions, on-device privacy, platform integration). It's not trying to beat Gemini at Gemini's strengths.

What about Liquid Glass UI?

iOS 27 also ships Liquid Glass, a new design language with glass-texture app icons and a continuous transparency slider. App icons now have a frosted-glass effect with depth. The transparency slider lets you adjust how translucent panels and toolbars are across the system.

This is purely cosmetic. Liquid Glass works on every iPhone that supports iOS 27 (iPhone 11 and later). It is not an Apple Intelligence feature.

What about privacy?

Apple's marketing is consistent on three layers:

On-device first — most tasks never leave your iPhone.
Private Cloud Compute second — when cloud is needed, Apple silicon processes your data with no retention. Apple has published the source for review.
Gemini third — when even Private Cloud Compute can't handle the task, Gemini is the fallback. Apple's deal contractually bars Google from using Apple user data for training.

The deal terms are strict: Google's Gemini processing of Apple user data happens in a sandboxed environment, and the output cannot be used to train future Gemini models. Apple has published the security architecture for review (similar to its Private Cloud Compute publication in 2024).

The honest critique: even with all three layers, Apple is now dependent on Google for ~30% of Apple Intelligence 2.0 tasks. If Google's Gemini terms change in 2027-2028, Apple would have 12-18 months to migrate. That's a real dependency-risk concern.

Should you update to iOS 27 on day one?

Yes, if you have iPhone 16 Pro or later. You get the full Apple Intelligence 2.0 stack, all seven features above work, and the on-device 3B model runs at full speed on the A18 Pro / A20 Pro NPU.

Yes, but with lower expectations, if you have iPhone 15 Pro / Pro Max. You get base Apple Intelligence 2.0, Visual Intelligence 2, and some Live Translation. The on-device 3B model runs in limited mode.

Maybe wait, if you have iPhone 15 or earlier. iOS 27 ships with Liquid Glass UI and the regular OS improvements, but you won't get Apple Intelligence 2.0. There's no urgency to update — iOS 26 was supported through iOS 26.7.

What's coming in iOS 27.1?

Five things to expect in October-November 2026 updates:

Five more Live Translation languages — Arabic, Hindi, Polish, Thai, Greek.
Group FaceTime Live Captions (currently 1:1 only).
Voice Memo transcription in Spanish, French, German.
Siri personal context for more third-party apps — currently Apple apps and a handful of major third-party apps (WhatsApp, Slack, Telegram).
Image Playground v2 — improved coherence, better text rendering, new "Photoreal" style.

These are scheduled for the iOS 27.1 (October), iOS 27.2 (November), and iOS 27.3 (December) updates.

Bottom line: Apple Intelligence 2.0 is the most significant iOS feature since iOS 13 (Dark Mode) and iOS 14 (Widgets). Siri 2.0 + Live Translation + Visual Intelligence 2 are the three features that change daily iPhone use. The Apple-Google Gemini partnership is real and consequential — Apple is no longer pretending to be fully independent on AI foundation models. Hardware gating is strict: iPhone 16 Pro or later is the floor for the full experience. Regional gating is real: EU and China get reduced features. For users in the U.S., Canada, U.K., India, Japan, Korea, and Australia with iPhone 16 Pro or later, iOS 27 on Monday Sept 14 is a must-update.