At WWDC 2025, Apple unveiled its Foundation Models framework, enabling developers to integrate Apple’s on-device AI models into their apps for enhanced features.
Apple highlighted that this framework lets developers utilize AI models without incurring inference costs. Additionally, these on-device models come equipped with features like guided generation and tool calling.
With iOS 26 now available to everyone, developers are updating their apps to leverage Apple’s on-device AI. Compared to the larger models from OpenAI, Anthropic, Google, or Meta, Apple’s models are relatively compact. As a result, these local features tend to enhance user experience rather than overhaul app workflows.
Here are some of the earliest apps making use of Apple’s AI framework.
Lil Artist
Lil Artist provides interactive activities to help children develop skills in creativity, math, and music. Developers Arima Jain and Aman Jain introduced an AI-powered story creator in the iOS 26 update. Users can pick a character and theme, and the app generates a story using AI. The developer confirmed that story generation relies on the local model.

Daylish
The creator of Daylish is developing a feature that automatically recommends emojis for timeline events in the daily planner app, based on the event’s title.
MoneyCoach
MoneyCoach, a finance management app, uses local models for two main features. It provides insights into your spending, such as highlighting if you spent more than usual on groceries that week. It also suggests categories and subcategories for expenses to speed up entry.

LookUp
LookUp, a vocabulary app, has introduced two new modes powered by Apple’s AI. One mode uses a local model to generate example sentences for words, prompting users to explain the word’s usage.

The developer also employs on-device models to create a map showing the origin of a word.

Tasks
Similar to other apps, Tasks now uses local models to automatically suggest tags for entries. It also detects recurring tasks and schedules them, and allows users to dictate items, which are then split into separate tasks using the on-device model, all without needing an internet connection.

Day One
Day One, the journaling app from Automattic, utilizes Apple’s models to highlight key points and suggest titles for entries. The app can also generate prompts to encourage users to expand on what they’ve written.

Crouton
Crouton, a recipe app, leverages Apple Intelligence to recommend tags for recipes and assign names to timers. It also uses AI to break down recipe instructions into clear, step-by-step directions.
Signeasy
Signeasy, a digital signature app, uses Apple’s on-device models to extract important details from contracts and provide users with a summary of the documents they’re signing.
Dark Noise
Dark Noise, an ambient sound app, allows users to describe a soundscape in a few words and then generates it using Apple’s local models. After creation, users can adjust the levels of different sound elements.
Lights Out
Lights Out, a new app for following the F1 season and grand prix by Shihab Mehboob (creator of Avery and Mammoth), uses on-device AI to summarize race commentary.
Capture
Capture, a note-taking app, employs local AI to suggest categories as users type notes or tasks.

Lumy
Lumy, an app for tracking sunlight and weather, now provides smart weather suggestions within the app using AI.

CardPointers
CardPointers helps users manage credit card spending and find the best ways to earn rewards. The latest version uses AI to answer questions about cards and available offers.

Guitar Wiz
Guitar Wiz, a guitar learning app, utilizes the Apple Foundation Model framework in several ways. It explains chords as users learn them, provides advanced insights based on practice intervals, and supports over 15 languages with the help of AI.
SmartGym
SmartGym uses on-device AI to turn workout descriptions into step-by-step routines with repetitions, intervals, and equipment details. It also generates workout summaries, monthly progress reports, routine breakdowns, and performance stats for each exercise.
Stoic
Stoic, a journaling app, leverages Apple’s models to provide users with personalized prompts based on mood tracking. The models also help summarize entries, search previous posts, and organize them.

SwingVision
SwingVision assists racquet sport players, including tennis and pickleball, in improving their technique through video analysis. The app now uses Foundation models to deliver targeted, actionable feedback.
Zoho
Zoho, an Indian productivity suite, employs on-device models for summarizing, translating, and transcribing content in its apps, such as Notebook for documents and Tables for spreadsheets.
TrainFitness
TrainFitness, a workout app, uses on-device models to recommend alternative exercises when specific equipment isn’t available.
Stuff
Stuff, a to-do list app, features a listen mode that uses Apple’s AI models to transcribe spoken input into individual tasks.
This list will be updated as we find more apps utilizing Apple’s on-device models.