AI Models

Apple May Allow Custom AI Models in iOS 27

Apple plans to let users select preferred third-party AI models to power Apple Intelligence features across iOS 27, iPadOS 27, and macOS 27 this fall. These AI extensions will support Siri, Writing Tools, and Image Playground. The company tests integrations with Google and Anthropic models, beyond the current ChatGPT support.

Neura News

Neura News

Neura Market Editorial

May 5, 20263 min read

Originally reported by theverge.com

Apple May Allow Custom AI Models in iOS 27

Apple May Allow Custom AI Models in iOS 27

Apple's upcoming operating system updates might give users the option to select their preferred AI model for Apple Intelligence features. Bloomberg reporter Mark Gurman states that the company intends to enable third-party chatbots to handle these AI functions throughout iOS 27, iPadOS 27, and macOS 27. All three updates should arrive this fall.

These third-party options, known as Extensions, would extend beyond Siri. They could also operate other Apple Intelligence tools, such as Writing Tools and Image Playground. Apple Intelligence, introduced as part of Apple's push into on-device AI processing, aims to enhance user tasks with privacy-focused computations. The suite debuted in earlier iOS versions and has grown with partnerships like the one with OpenAI for ChatGPT integration.

Custom Siri Voices for Different Models

Gurman notes that Apple will permit users to assign distinct Siri voices to various AI models. For example, Siri powered by Apple's own models might use one voice. Meanwhile, Siri running on ChatGPT could employ a different one. This flexibility builds on Siri's ongoing upgrades. Apple recently announced plans to overhaul Siri using Google's AI models as a foundation, marking a shift from its in-house development alone.

Users would pick from AI providers that choose to participate. Developers add support through their App Store applications. After installing a compatible app, people set it as the default AI option in the Settings app. This setup mirrors how users already manage default apps for browsers or email.

Internal Testing with Google and Anthropic

Apple has begun internal tests of integrations with AI models from Google and Anthropic. Google's models form the core of the forthcoming Siri improvements. Anthropic, known for its Claude family of AI models, represents another potential partner. Claude models emphasize safety and helpfulness, competing directly with offerings from OpenAI and Google.

The #1 Newsletter in AI

Stay ahead of the AI curve

The most important updates, news, and content — delivered weekly.

No spam. Unsubscribe anytime.

To date, ChatGPT stands as the sole third-party model linked to Apple Intelligence. That arrangement started last year, allowing on-device processing for privacy while tapping OpenAI's capabilities for complex queries. ChatGPT, developed by OpenAI, powers tools like image generation and text completion in Apple's ecosystem. The expansion to more providers could broaden choices and spur competition among AI firms.

Mark Gurman, a respected analyst at Bloomberg with a track record on Apple scoops, often reveals details ahead of official announcements. His reports have accurately predicted features in past iOS releases. Apple has not confirmed these plans publicly, but the timeline aligns with the company's annual fall software cycle.

This development comes amid intense rivalry in AI. Apple trails rivals like Google and OpenAI in generative AI but focuses on seamless integration into daily devices. iPhones, iPads, and Macs number in the billions worldwide. Allowing Extensions could keep Apple competitive by letting users access the best models without switching ecosystems.

The changes target system-wide use, meaning AI assistance feels native regardless of the underlying model. Providers must meet Apple's standards for App Store inclusion, ensuring quality and security. Fall releases typically accompany new hardware like iPhones, potentially showcasing these features at events in September.

Related on Neura Market

More from Neura News

AI Models

Google Unveils Gemini 3.6 Flash, 3.5 Flash-Lite, and Cyber Model

Google has released three new Gemini models: 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber. The 3.6 Flash model offers improved coding and knowledge work with 17% fewer output tokens and lower costs. The 3.5 Flash-Lite is the fastest in the series at 350 tokens per second, designed for high-throughput agentic tasks. The 3.5 Flash Cyber model, available only to governments and trusted partners via CodeMender, focuses on finding and fixing cybersecurity vulnerabilities. Google also noted that Gemini 3.5 Pro is being tested with partners and that pre-training for Gemini 4 has begun.

Jul 21·5 min read