How developers use Apple's local AI model on iOS26

Applications of AI


Earlier this year, Apple introduced the basic model framework in WWDC 2025. This allows developers to use their local AI model to enhance their application capabilities.

The company touted that with this framework, developers can access AI models without worrying about inference costs. Additionally, these local models have features such as guided generation and built-in tool calls.

With iOS 26 deployed to all users, developers are updating their apps to include features powered by Apple's local AI model. Apple's models are smaller compared to major models from Openai, Anthropic, Google, or Meta. So, local only features significantly improve the quality of life with these apps, rather than introducing major changes to the app's workflow.

Below are some of the first apps to leverage Apple's AI framework:

Lil Artist

The LIL Artist App offers a variety of interactive experiences to help children learn different skills such as creativity, mathematics, music and more. Developer Arima Jain has shipped AI Story Creators with iOS 26 update. This allows users to select characters and themes, and the app uses AI to generate a story. The developer said that story text generation is equipped with a local model.

Image credits:Lil Artist

Daylish

The developers of the Daylish app are working on a prototype to automatically propose emojis for timeline events based on the title of the Daily Planner app.

Moneycoach

The finance tracking app MoneyCoach has two neat features, featuring a local model. First, the app provides insight into spending, such as whether you spent above average on groceries for that particular week. Another feature automatically suggests categories and subcategories of spend items for quick entry.

Screenshots from the financial tracking app MoneyCoach, where the screen outlines your account and shows weekly insights into grocery spending for the week.
Image credits:Moneycoach

look up

Word Learning App Lookup has added two new modes using Apple's AI model. There is a new learning mode, which utilizes local models to create examples that correspond to words. Additionally, this example asks the user to explain the use of words in a sentence.

Screenshot of lookup word learning app showing screen showing examples for understanding new words
Image credits:look up

Developers are also using models on the device to generate map views of the origin of words.

Screenshot of lookup app with a map view showing the origin of the word.
Image credits:look up

task

Like some other apps, the Tasks app implemented the ability to automatically use the local model to suggest tags for entries. They also use these models to detect recurring tasks and schedule them accordingly. Also, with the app, users can talk about a few things and break them down into different tasks without using the internet using the local model.

This is a lifestyle shot of a task app showing a screen with a new Feauutre that suggests tags when entering tasks using a local model.
Image credits:task

first day

The first day of the journaling app owned by Automattic uses Apple's models to get highlights and propose the title of the entry. The team has also implemented a feature to generate prompts to dive deeper and tweak you to write more based on what you already wrote.

A screenshot showing the functionality of the first day journaling app implemented using Apple's local model. These features include summaries, title suggestions, and rapid generation deepening into writing.
Image credits:first day

Crutons

The recipe app Crouton uses Apple Intelligence to suggest recipe tags and assign names to timers. It also uses AI to break down blocks of text into simple follow steps for cooking.

A subtle

Digital Signasis app Signeasy uses Apple's local model to extract key insights from contracts and provide a summary of the documents that users are signing.

We will continue to update this list to discover more apps using Apple's local models.



Source link

Leave a Reply

Your email address will not be published. Required fields are marked *