Apple May Let You Pick Your AI Model in iOS 27

Apple is reportedly planning a dramatic shift in its AI strategy that would allow users to choose their preferred artificial intelligence model to power Apple Intelligence across the entire operating system. According to Bloomberg's Mark Gurman, one of the most reliable Apple insiders, the company is preparing to open up iOS 27, iPadOS 27, and macOS 27 to third-party chatbots — a move that could fundamentally reshape how over 1 billion Apple device owners interact with AI.

The change, expected to arrive this fall alongside Apple's annual software updates, represents a significant departure from the company's traditionally closed-ecosystem approach. Rather than forcing users to rely solely on Apple's in-house AI capabilities, the update would let competing models like OpenAI's ChatGPT, Google's Gemini, and potentially Anthropic's Claude serve as the default intelligence layer powering Siri and other system-wide AI features.

Key Takeaways

• User choice: Apple plans to let users select their preferred AI model to run Apple Intelligence features system-wide
• Fall 2025 timeline: The feature is expected in iOS 27, iPadOS 27, and macOS 27
• Third-party integration: External chatbots could power core OS features, not just standalone apps
• Strategic pivot: Apple moves from a walled-garden AI approach to an open-marketplace model
• Competitive pressure: The shift comes as Apple Intelligence has faced criticism for lagging behind rivals
• Source credibility: Bloomberg's Mark Gurman, who has an exceptional track record on Apple leaks, broke the story

Apple Acknowledges Its AI Limitations

This potential move signals something remarkable — Apple appears to be acknowledging that its own AI models may not be best-in-class. When the company launched Apple Intelligence at WWDC 2024, it positioned the feature as a cornerstone of its device ecosystem. Yet the rollout has been rocky, with critics pointing out that Siri's AI-powered upgrades still trail behind competitors like ChatGPT and Google Gemini in terms of capability and reliability.

Apple's on-device AI models, while impressive for privacy-focused tasks, have struggled with the kind of complex reasoning and creative generation that cloud-based models from OpenAI and Google handle with ease. The company already took a half-step toward openness last year by integrating ChatGPT as an optional fallback within Siri, but this new plan goes much further.

By letting users swap out the underlying AI engine entirely, Apple is essentially admitting that a one-size-fits-all approach to AI doesn't work. This is a pragmatic concession from a company that has historically prided itself on controlling every layer of the user experience.

How the Model Marketplace Could Work

While specific implementation details remain scarce, Gurman's report suggests that the third-party AI integration would go far beyond what Apple currently offers. Today, Siri can hand off certain queries to ChatGPT, but the interaction feels bolted on rather than seamless. In iOS 27, the chosen AI model could potentially power:

• Writing tools across Mail, Notes, and Messages
• Siri conversations and follow-up queries
• Image generation and editing in Photos
• Summarization features in Safari and other apps
• Smart Reply suggestions across the system
• On-screen awareness and contextual understanding

This would effectively turn Apple's operating system into a platform-agnostic AI shell, where the intelligence layer is modular and user-selected. Think of it like choosing a default browser or email app — except the stakes are far higher because AI touches virtually every interaction on the device.

The technical challenge here is significant. Each AI model has different capabilities, API structures, and response formats. Apple would need to build a standardized interface layer that translates system requests into model-specific calls while maintaining a consistent user experience regardless of which model is running underneath.

The Competitive Landscape Driving Apple's Decision

Google has been aggressively integrating its Gemini models across Android, with Gemini replacing Google Assistant as the default AI on Pixel and Samsung devices. Samsung's Galaxy AI, powered by a combination of Google and Samsung's own models, has also raised the bar for on-device AI features. Meanwhile, Microsoft continues to embed Copilot deeply into Windows 11 and its productivity suite.

Apple risks falling behind if it clings to a proprietary-only AI strategy. The numbers tell the story: OpenAI's ChatGPT has surpassed 400 million weekly active users as of early 2025, while Google Gemini is rapidly gaining ground. Users are developing strong preferences for specific AI models, and forcing them to use Apple's comparatively limited offering could drive dissatisfaction — or worse, platform switching.

The AI model market is also evolving at breakneck speed. New models from Meta (Llama 4), Anthropic (Claude 4), Mistral, and xAI (Grok) are launching every few months, each bringing unique strengths. By building a pluggable architecture, Apple positions itself to benefit from this innovation without having to match every competitor's R&D spending dollar for dollar.

Privacy Concerns and Apple's Balancing Act

One of the biggest questions surrounding this plan is how Apple will reconcile third-party AI integration with its long-standing commitment to user privacy. Apple Intelligence was designed with privacy as a core selling point — processing data on-device whenever possible and using Private Cloud Compute for more demanding tasks.

Routing user data through third-party AI models inherently introduces privacy trade-offs. When a query goes to OpenAI's servers or Google's infrastructure, Apple loses direct control over how that data is processed, stored, and potentially used for model training. The company will need to establish strict data-handling requirements for any AI provider that wants to participate in the program.

Apple could potentially address this through several mechanisms:

• Mandatory data deletion policies for third-party providers
• On-device preprocessing to strip personally identifiable information before sending queries
• Transparency labels showing users exactly what data each model accesses
• Tiered permissions letting users control which features use cloud-based AI vs. on-device processing

This privacy framework could actually become a competitive advantage — positioning Apple as the trusted gatekeeper that ensures AI providers meet strict standards before accessing user data.

What This Means for Developers and AI Companies

For AI companies, access to Apple's ecosystem represents an enormous opportunity. With over 1.5 billion active Apple devices worldwide, becoming an available AI model option in iOS 27 could drive massive user adoption. The revenue implications are substantial — if Apple takes a cut of subscription fees for premium AI models (similar to its App Store model), it could create an entirely new revenue stream.

For app developers, the implications are equally significant. If the system-wide AI model changes, apps that rely on Apple Intelligence features may need to account for varying capabilities across different models. A writing assistant that works perfectly with Claude might behave differently when the user switches to Gemini. Apple will likely need to provide robust developer tools and APIs to handle these variations gracefully.

The move could also accelerate AI model competition in unexpected ways. If users can directly compare how different models perform within the same familiar interface, it creates a transparent benchmarking environment. Models that underperform in real-world iOS tasks will quickly lose users to superior alternatives, driving faster innovation.

Looking Ahead: The Fall 2025 Timeline and Beyond

Apple is expected to unveil iOS 27 at WWDC 2025 in June, with a public release likely in September alongside new iPhone hardware. If the third-party AI model feature makes the cut, it will almost certainly be one of the headline announcements at the keynote.

However, the feature may launch with a limited set of partner models. Apple is known for carefully curating its ecosystem, and the company will likely start with 2-3 established providers — most probably OpenAI, Google, and possibly Anthropic — before expanding access to smaller AI companies.

The broader significance of this move extends beyond Apple. If the world's most influential consumer technology company embraces an open AI marketplace model, it could set a precedent for the entire industry. We may see similar approaches from other platform holders, creating a future where AI models compete on merit within standardized platforms rather than being locked to specific ecosystems.

For now, Apple users should watch WWDC 2025 closely. The era of choosing your AI — the way you choose your browser or search engine — may be just months away. And that level of consumer choice could be exactly what the AI industry needs to push models from 'good enough' to truly exceptional.