Apple Intelligence, Galaxy AI and Gemini on Pixel all put generative AI into the phone, yet each vendor documents devices, features, cloud processing and privacy in different places and different words, which makes a straight comparison hard to assemble. This page lays the official facts for the three side by side: what each is, which devices and operating systems run it, which features are named, how on-device and cloud processing are split, what the privacy architecture says, what is free and what may become paid, how AI-edited images are labelled and what developers can build on. It is written for everyday buyers, IT and mobile device teams deciding what staff may use, and privacy teams answering "where does my data go?". Platforms are listed alphabetically (Apple Intelligence, Galaxy AI, Gemini on Pixel) everywhere on the page and order is never a ranking.
Last checked: 2026-11-15.
This is an independent reference page from NHI Mgmt Group based on public provider documentation; it is not sponsored or endorsed by any provider. Device lists, features and plans change often, so confirm them on each provider's current pages.
Key takeaways
- Three different shapes. Apple Intelligence is a system feature across iPhone, iPad, Mac, Vision Pro and Apple Watch; Galaxy AI is Samsung's feature suite on Galaxy phones, tablets and wearables; Gemini on Pixel combines Google's on-device Gemini Nano model with Gemini cloud models on Pixel phones.
- Device floors are documented differently. Apple lists iPhone 15 Pro, iPhone 16 and later plus M1 or later iPad and Mac; Samsung lists Galaxy S24 series and later phones; Google's ML Kit page lists Pixel 9, 10 and 11 series for Gemini Nano APIs, with the Gemini Nano version depending on the device and its RAM.
- Each names a private cloud. Apple uses Private Cloud Compute on Apple silicon servers; Google announced Private AI Compute on its TPUs and Titanium Intelligence Enclaves; Samsung describes a Personal Data Engine with KEEP and Knox Vault on the device and a user choice between local and cloud processing.
- Free today, fees signposted. Apple says expanded access to server features will be available for a fee in the future; Samsung says basic Galaxy AI features are free and future advanced features may be paid; Google bundles Google AI Pro trials with Pixel Pro models.
- Labelling differs by tool. Samsung overlays a visible watermark on images saved from Photo Assist; Google adds C2PA Content Credentials to every Pixel 10 camera photo and to AI-edited images in Google Photos; Apple's Apple Intelligence support page does not describe a label.
What are Apple Intelligence, Galaxy AI and Gemini on Pixel?
Apple Intelligence is Apple's personal intelligence system built into iOS, iPadOS, macOS, visionOS and watchOS. Apple's support page (article 121115) lists features from Writing Tools and notification summaries to Image Playground, Genmoji, Live Translation and a ChatGPT extension. Larger requests run on Private Cloud Compute, which Apple's security blog describes as a cloud system for private AI processing that lets Apple Intelligence use larger foundation models.
Galaxy AI is Samsung's suite of AI features on Galaxy devices. Samsung's Galaxy AI page names features such as Now Nudge, Now Brief, Photo Assist, Creative Studio, Audio Eraser, Call Screening, Transcript Assist, Interpreter, Writing Assist and Bixby. Galaxy phones also ship with Google's Gemini and, on the Galaxy S26 series, a pre-installed Perplexity app, which Samsung lists alongside Bixby as part of its AI ecosystem.
Gemini on Pixel means the AI features on Google Pixel phones, built on Gemini Nano running on the device through Android's AICore system service, plus Gemini cloud models. Google's Android developer page describes Gemini Nano as its on-device foundation model that lets apps run generative AI without a network connection or sending data to the cloud. Pixel features such as Magic Cue use it; Google's Pixel 10 announcement says Tensor G5 and the latest Gemini Nano run Magic Cue on the phone.
Single-product depth is on Apple Intelligence Explained, Galaxy AI Explained and Gemini Nano Explained; this page keeps to the side-by-side view.
The three platforms at a glance
As of 2026-11-15, from each provider's own pages.
| Field | Apple Intelligence | Galaxy AI | Gemini on Pixel |
|---|---|---|---|
| Provider | Apple | Samsung | |
| Product form | System feature across Apple operating systems | Feature suite on Galaxy devices (One UI) | Pixel features built on Gemini Nano and Gemini cloud models |
| On-device model or engine named | Apple Intelligence models and AFM 3 Cloud models, both named for Shortcuts | Personal Data Engine for personal data processing | Gemini Nano in AICore (versions 2, 3 and 4 listed by device) |
| Private cloud named | Private Cloud Compute | User choice of local or cloud processing in privacy settings | Private AI Compute (announced 11 Nov 2025) |
| Latest flagship cited | iOS 27 generation devices including iPhone Duo and iPhone Air | Galaxy S26 series, unveiled 25 Feb 2026 | Pixel 11 series, announced 12 Aug 2026, Tensor G6 |
| Third-party assistant integration | ChatGPT extension, where the ChatGPT app and service is available | Gemini (Google Account, users 18+) and Perplexity (pre-installed on S26) | Gemini is Google's own assistant |
| Price statement | Daily limits on server features; expanded access for a fee in the future | Basic features free; future advanced features may be paid | Pixel 11 Pro models include six months of Google AI Pro (US) |
| Image labelling | Not stated by Apple on the support page checked | Visible watermark on images saved from Photo Assist | C2PA Content Credentials on Pixel 10 camera photos and AI-edited Google Photos images |
Which devices support each one?
Each provider publishes its own device list. The table below quotes the lists as they appear on the official pages; regional and carrier variations apply to all three.
| Field (as of 2026-11-15) | Apple Intelligence | Galaxy AI | Gemini on Pixel |
|---|---|---|---|
| Phones | iPhone Duo; iPhone 16 models or later; iPhone 15 Pro and 15 Pro Max; iPhone Air | Galaxy S24 series and later, plus foldables such as Z Fold8 and Z Fold8 Ultra | Pixel 9, Pixel 10 and Pixel 11 series (ML Kit GenAI list) |
| Tablets and computers | iPad mini (A17 Pro and A20 Pro); iPad with M1 or later; Mac with M1 or later; MacBook Neo (A18 Pro) | Tablets such as Tab S11 Ultra, Tab S11 and Tab S10+; support depends on model and OS version | Not part of the Pixel phone feature set; ML Kit lists Googlebook laptops from several makers for Gemini Nano |
| Wearables and other | Apple Vision Pro; Apple Watch Series 9 or later, Ultra 2 or later and SE 3 with a paired Apple Intelligence iPhone | Galaxy Watch and Galaxy Buds models, for example Watch Ultra2, Watch9, Buds4 Pro and Buds4 | Not stated by Google for Gemini Nano wearables on the pages checked |
| Operating system | iOS 27, iPadOS 27, macOS 27 Golden Gate, visionOS 27, watchOS 27; some features from the 26 releases | One UI; Knox Matrix needs One UI 7.0 or later and KEEP needs One UI 8 or later | Android with AICore; Gemini Nano version depends on OS version and AICore update status |
| Storage or memory note | Up to 14 GB storage on newer models, up to 8 GB on others, up to 1.5 GB on Apple Watch | Not stated by Samsung on the Galaxy AI page | Model availability and version may vary with the device's RAM configuration |
Note on scope: Gemini Nano also runs on many non-Pixel Android phones, including the Galaxy S25 and S26 series, Honor, Motorola, OnePlus, OPPO, Xiaomi and others named on the ML Kit page. That is a developer-facing model list, not the set of Pixel-branded features. See the Gemini Nano Explained page for the full tier list.
What features does each platform name?
Feature names differ even where the task is similar. The table groups features by task; each cell lists only names that the provider's own page uses.
| Task (as of 2026-11-15) | Apple Intelligence | Galaxy AI | Gemini on Pixel |
|---|---|---|---|
| Writing help | Writing Tools; automatic proofreading; Mail and Messages smart reply | Writing Assist (proofread, rewrite, summarise) | Gemini Nano text APIs used by apps: summarisation, proofreading, rewriting |
| Summaries and transcripts | Mail, message, notification, voicemail and notes summaries | Transcript Assist; Now Brief | Pixel Recorder summaries (named for Private AI Compute) |
| Contextual suggestions | Call Context; natural-language calendar and reminder creation | Now Nudge | Magic Cue in Google Messages and the Phone app |
| Photo editing | Clean Up, Extend and Spatial Reframing; Image Playground; Genmoji; Image Wand | Photo Assist; Creative Studio; Audio Eraser; Auto Trim; My FanCam | Camera Coach; Pro Res Zoom on Pixel 10 Pro models; imaging models on Tensor G6 |
| Translation and calls | Live Translation in Messages, Phone, FaceTime and AirPods | Interpreter; Call Screening | Speech Recognition, Speech Translation and Text to Speech APIs in ML Kit |
| Assistant | Siri AI (a server-side feature); ChatGPT extension | Bixby; Gemini; Perplexity | Gemini; Gemini Intelligence named for Pixel 11 |
Requirements written into footnotes: Samsung marks several features as needing a network connection and a Samsung Account (Photo Assist, Creative Studio, Transcript Assist and Writing Assist among them). Apple lists Siri AI, Intelligent Photo Editing Tools, Image Playground and AFM 3 Cloud models in Shortcuts as server-side features with daily usage limits. Google's ML Kit page permits Gemini Nano inference only when the calling app is the top foreground app and enforces a per-app quota through AICore.
How is work split between the device and the cloud?
Apple Intelligence: requests run on the device where possible and larger ones go to Private Cloud Compute. Apple's support page names the server-side features subject to daily limits, which is the clearest public list of features that reach the cloud. The ChatGPT extension is a separate route to OpenAI's service and is available only where the ChatGPT app and service is available.
Galaxy AI: Samsung's Galaxy AI page says users can choose whether Galaxy AI data is processed locally on the device or in the cloud through privacy settings. Personal data handled by the Personal Data Engine is processed on the device, then encrypted and stored with KEEP and Knox Vault; the engine works only while Personal Data Intelligence is turned on, and analysed data is deleted when it is turned off. Features footnoted as needing a network connection use cloud processing.
Gemini on Pixel: Gemini Nano prompts run on the device with no server calls, according to the Android developer page. When Gemini Nano is unavailable or a request exceeds its limits, developers can route requests to Gemini Flash or Gemini Pro in the cloud through the Firebase AI Logic hybrid API. For Google's own features, Private AI Compute handles cloud requests for Magic Cue suggestions and Recorder summaries in more languages.
The trade-offs between these models are covered in On-Device vs Cloud AI.
What does each provider say about privacy?
All three describe hardware-backed protections. The wording below follows each provider's own documentation; none of it is an independent assessment.
| Privacy field (as of 2026-11-15) | Apple Intelligence | Galaxy AI | Gemini on Pixel |
|---|---|---|---|
| Private cloud and its hardware | Private Cloud Compute on custom Apple silicon servers with Secure Enclave and Secure Boot | Cloud processing optional through privacy settings | Private AI Compute on Google TPUs with Titanium Intelligence Enclaves |
| Retention statement | Data used only to fulfil the request; no user data retained after the response is returned | Analysed Personal Data Engine data deleted when Personal Data Intelligence is turned off | AICore does not store inputs or outputs after processing |
| Provider access statement | Not accessible to anyone, including Apple staff; no remote shell or interactive debugging on nodes | Personal data encrypted and stored with KEEP and Knox Vault on the device | Data in Private AI Compute is not accessible to anyone else, not even Google, per Google |
| Verification mechanism | Published PCC software images and a transparency log; devices send data only to attested nodes | Knox Matrix Trust Chain, where Galaxy devices check on one another | Remote attestation and encryption link the device to a sealed cloud environment |
| On-device isolation | Not described on the support page checked | KEEP and Knox Vault storage | AICore isolated from most packages; no direct internet access, downloads go through Private Compute Services |
For enterprise controls on Apple devices, see Apple Intelligence Enterprise Security. For the wider security picture of models running on phones and laptops, see the On-Device AI Security Guide.
Languages and regions
Apple Intelligence: on iOS, iPadOS and macOS, Apple lists English, Danish, Dutch, French, German, Italian, Norwegian, Portuguese, Spanish, Swedish, Turkish, Vietnamese, Chinese (simplified and traditional), Japanese and Korean. The device language and Siri language must be set to the same supported language. Apple says Apple Intelligence features will not currently work for supported devices purchased in mainland China, and lists specific features now available to EU users.
Galaxy AI: Samsung says feature availability varies by region, model, OS version and carrier, and service availability and accuracy vary by language. Interpreter may need language pack downloads, and Bixby recognises only certain accents and dialects.
Gemini on Pixel: Google's ML Kit page says language support may vary with the device's configuration and downloaded models. The Pixel 11 Google AI Pro offer is US-only according to the announcement footnotes. Gemini on Galaxy phones is limited to select languages and users aged 18 and over, per Samsung's S26 footnotes.
Pricing and plans
| Field (as of 2026-11-15) | Apple Intelligence | Galaxy AI | Gemini on Pixel |
|---|---|---|---|
| Current charge for features | Included with supported devices; server features have daily usage limits | Basic Galaxy AI features provided by Samsung are free | Pixel features included; Gemini plans sold separately |
| Future paid tier statement | "Expanded access to such features will be available for a fee in the future" | Future releases may include advanced features in a paid subscription plan | Not stated by Google for Pixel features on the pages checked |
| Bundled subscription | HomeKit Secure Video features need an eligible iCloud+ plan | Third-party AI features may have different terms | Pixel 11 Pro and Pro XL: six months of Google AI Pro (US, post-trial US$19.99 a month) |
| Flagship launch price cited | Not used here (feature spans many devices) | Galaxy S26 from £879 (UK RRP, 256GB) | Pixel 11 from US$899 (US pre-order, 256GB) |
Provenance and labelling of AI images
Apple Intelligence: Apple's Apple Intelligence support page lists Clean Up, Extend, Spatial Reframing and Image Playground but does not describe how edited or generated images are labelled. Check Apple's Photos documentation for current behaviour.
Galaxy AI: Samsung says a visible watermark is overlaid on images saved from Photo Assist to indicate they were generated by Galaxy AI. The S26 article does not state a watermark for Creative Studio outputs.
Gemini on Pixel: Google's security blog (10 September 2025) says the Pixel 10 lineup adds C2PA Content Credentials to every Pixel Camera JPEG photo, and Google Photos adds them to images edited with AI tools. The Pixel Camera app achieved Assurance Level 2 in the C2PA Conformance Program.
Visible watermarks and signed metadata solve different problems: the first is readable by anyone but easy to crop, the second survives intact files and can be verified but is lost when platforms strip metadata. See the AI Content Provenance and C2PA Guide and SynthID vs C2PA.
What can developers build on?
Apple Intelligence: Apple's support page lists "Use Apple Intelligence models in Shortcuts" and AFM 3 Cloud models in Shortcuts, subject to daily limits for server features. Developer frameworks are documented on Apple's developer site and covered in Apple Intelligence Explained.
Galaxy AI: Samsung's consumer Galaxy AI page does not describe a developer API. Galaxy S25 and S26 devices are listed by Google as Gemini Nano 4 devices, so Android apps on those phones can call Gemini Nano through ML Kit GenAI.
Gemini on Pixel: ML Kit GenAI exposes GenerativeModel for direct prompts, AgentRunner for multi-step tool calling, and task APIs for image description, summarisation, proofreading, rewriting, semantic search and speech. Apps should call getBaseModelName() to see which Gemini Nano version a device runs, since output may differ between versions.
How to choose: factual criteria
There is no single right answer; the choice usually follows the device estate already in place. These factual questions help:
- Which devices do you already manage? Apple Intelligence needs Apple hardware from the lists above; Galaxy AI needs Galaxy S24 series or later; Pixel features need a Pixel phone. Gemini Nano is available to apps on many Android brands.
- Where may data be processed? Compare the named private cloud for each platform with your data residency and retention rules, and check whether Samsung's local-processing setting suits sensitive roles.
- Which languages and countries matter? Apple's language list and mainland China restriction, Samsung's per-region availability and Google's per-device language support all differ.
- Which third-party assistants are acceptable? Apple's ChatGPT extension, Samsung's Gemini and Perplexity integrations and Google's own Gemini each carry their own terms.
- Do you need image provenance? Decide whether visible watermarks or C2PA Content Credentials meet your content policy.
- What happens when free access changes? All three providers signal paid tiers or subscriptions; budget for that before relying on server features.
Common failure patterns
- Treating "on-device" as a whole-product claim: every platform sends some requests to a cloud, and third-party assistants route to their own services.
- Assuming one device list: Gemini Nano support on a non-Pixel phone does not mean Pixel features are present, and Galaxy AI support depends on model, OS and carrier.
- Ignoring language settings: Apple requires matching device and Siri languages; staff on unsupported languages see fewer features.
- Relying on visible watermarks: a cropped watermark leaves no trace; metadata-based credentials are lost when files are re-encoded.
- Personal accounts on work phones: several Galaxy AI features need a Samsung Account and Gemini needs a Google Account, which ties AI use to personal identities unless managed accounts are enforced.
Where identity and security fit
Account requirements make these features identity decisions as much as device decisions. Apple ties features to the Apple Account and its country or region; Samsung footnotes Samsung Account login for many features; Google features need a Google Account. Managed devices should use organisation-controlled accounts where the platform allows it, and mobile device management policy should state which AI features and third-party assistants are permitted. See the BYOD Security Programme and Block vs Sanction vs Coach for the policy side.
Practitioner checklist
- Inventory which phones, tablets and laptops meet each provider's published device list.
- Record which features on each platform reach a cloud service, using the providers' own server-feature lists and network-connection footnotes.
- Decide whether Samsung's local-processing setting and Apple's ChatGPT extension are allowed on managed devices.
- Require organisation-managed Apple, Samsung and Google accounts on work devices where supported.
- Set a content policy for AI-edited images covering visible watermarks and C2PA Content Credentials.
- Check language and region limits for every country where staff work.
- Track each provider's paid-tier announcements before depending on server features.
- Recheck device lists and features each quarter, since all three update with new OS releases.
Standards and references
- Apple Support: How to get Apple Intelligence
- Apple Security Research: Private Cloud Compute
- Samsung UK: Galaxy AI
- Samsung Newsroom UK: Galaxy Unpacked 2026, Galaxy S26 series
- Android Developers: Gemini Nano
- Google ML Kit: GenAI APIs and supported devices
- Google: Private AI Compute
- Google: The Pixel 11 series
- Google: Pixel 10 Pro and Pixel 10 Pro XL
- Google Security Blog: Pixel C2PA Content Credentials
Related NHI Mgmt Group resources: On-Device AI and AI PCs Hub · Apple Intelligence Explained · Galaxy AI Explained · Gemini Nano Explained · On-Device AI Security Guide