Apple May Let Users Pick Any AI Model in iOS 27

Apple is preparing to let users choose their own preferred AI model to power Apple Intelligence, moving beyond its current exclusive arrangement with OpenAI's ChatGPT. The change, expected to arrive with iOS 27 this fall, would mark a dramatic shift in how the iPhone maker approaches artificial intelligence integration across its ecosystem.

According to a report from Bloomberg's Mark Gurman, Apple is developing system-level interfaces for iOS 27, iPadOS 27, and macOS 27 that would allow external chatbots to plug into and drive multiple Apple AI features. The move signals Apple's recognition that a single-provider AI strategy may not serve its massive global user base.

Key Takeaways

• Apple plans to open Apple Intelligence to multiple third-party AI models in iOS 27, iPadOS 27, and macOS 27
• Users will be able to select their preferred AI chatbot, not just ChatGPT
• System-level APIs are being developed to let external models power core Apple AI features
• The changes are expected to ship this fall alongside Apple's next major OS releases
• The shift represents a platform-style approach rather than a single-vendor partnership
• Bloomberg's Mark Gurman broke the story, citing people familiar with the matter

Apple Moves From Exclusive Partner to Open Platform

When Apple first introduced Apple Intelligence at WWDC 2024, it leaned heavily on a partnership with OpenAI to handle complex queries that its on-device models couldn't manage. ChatGPT became the default — and only — third-party option for Siri's extended capabilities, text generation, and image understanding tasks.

That exclusive arrangement drew both praise and criticism. Privacy advocates appreciated Apple's careful gatekeeping, while competitors like Google, Anthropic, and Meta were effectively shut out of the world's most lucrative mobile ecosystem. With iOS 27, Apple appears ready to change the equation entirely.

The planned system-level interface would function as a standardized bridge between Apple's native AI features and external large language models. This means users could potentially swap ChatGPT for Google Gemini, Anthropic's Claude, or other models — all while maintaining the same seamless Apple Intelligence experience on the surface.

Why Apple Is Opening the Door Now

Several factors are likely driving Apple's decision to embrace a multi-model approach. The AI landscape has evolved rapidly since Apple first announced its ChatGPT integration, and locking users into a single provider increasingly looks like a competitive disadvantage.

Market pressure plays a significant role. Samsung and Google have both pursued more flexible AI strategies on Android, with Samsung offering multiple AI backends and Google deeply integrating Gemini across its hardware lineup. Apple risks appearing inflexible if it continues to limit user choice.

There are also regulatory considerations. The European Union's Digital Markets Act and similar legislation worldwide push platform operators toward interoperability and user choice. By proactively offering model selection, Apple gets ahead of potential regulatory mandates that could force the change on less favorable terms.

Perhaps most importantly, the AI model market is fragmenting. Different models excel at different tasks — Claude is widely praised for long-form writing and analysis, Gemini integrates deeply with Google services, and specialized models from companies like Mistral and Perplexity offer unique capabilities. A one-size-fits-all approach no longer makes strategic sense.

How the System-Level API Could Work

While Apple hasn't publicly confirmed the technical architecture, the concept of a system-level interface for AI models suggests a carefully controlled integration framework. Based on Apple's historical approach to third-party integrations, several design principles are likely:

• Standardized input/output protocols that ensure consistent behavior regardless of which model is selected
• Privacy-preserving data handling that routes queries through Apple's infrastructure before reaching third-party models
• Performance benchmarks that third-party models must meet to qualify for integration
• User consent flows that clearly explain what data is shared with each provider
• Fallback mechanisms that switch to on-device processing if the selected cloud model is unavailable

This approach mirrors how Apple handles default browser and mail app selection in iOS — giving users choice while maintaining a consistent, Apple-quality experience. The key difference is the complexity involved in routing AI queries to different backends, each with unique capabilities, pricing structures, and API designs.

Apple's on-device models would likely continue handling simpler tasks like text summarization and notification prioritization. The third-party model selection would primarily affect complex queries routed through Siri, writing assistance features, and advanced image or document analysis.

The Competitive Implications Are Enormous

If Apple follows through on this plan, the ripple effects across the AI industry could be substantial. Access to Apple's ecosystem — with over 1.5 billion active devices worldwide — represents an enormous distribution channel for AI model providers.

For OpenAI, this is a double-edged sword. While ChatGPT would remain an option, it would lose its privileged position as the sole third-party AI provider on Apple devices. The company would need to compete on merit rather than exclusivity, potentially affecting its premium pricing strategy and enterprise negotiations.

For Google and Anthropic, the opportunity is massive. Google's Gemini could gain direct access to iPhone users without requiring them to download a separate app. Anthropic's Claude — already popular among developers and professionals — could reach a mainstream consumer audience for the first time through Siri integration.

Smaller players like Mistral, Cohere, and emerging open-source model providers could also benefit, assuming Apple's qualification requirements aren't prohibitively strict. This democratization of access could reshape how AI companies compete for consumer mindshare.

What This Means for Users, Developers, and Businesses

The practical implications of Apple's multi-model approach extend across multiple stakeholder groups:

For everyday users, the change means more control over their AI experience. Someone who prefers Claude's writing style could use it for email drafts, while another user might choose Gemini for its superior Google Workspace integration. The key question is how Apple presents these choices — whether through a simple settings toggle or a more nuanced per-feature selection.

For developers, system-level AI APIs could unlock powerful new capabilities. App makers could potentially leverage whatever model the user has selected, reducing the need to bundle their own AI backends. This could lower development costs and improve consistency across the Apple ecosystem.

For businesses and enterprises, model choice at the OS level addresses a critical concern. Companies with data residency requirements or existing AI vendor relationships could standardize their preferred model across corporate-managed Apple devices. This makes Apple hardware more attractive for enterprise AI deployments compared to locked-in alternatives.

The subscription model also comes into play. Currently, accessing ChatGPT through Apple Intelligence is free for basic queries, with premium features requiring a ChatGPT Plus subscription at $20 per month. A multi-model approach would need to address how billing works — whether through Apple's App Store payment system, direct subscriptions with each provider, or a bundled Apple One tier.

Looking Ahead: The Fall 2025 Timeline

Apple typically previews its next-generation operating systems at WWDC, historically held in June. If iOS 27 is indeed the version that introduces multi-model AI support, developers and AI companies should expect announcements at WWDC 2025 with a public release in September or October.

The timeline raises several questions that remain unanswered:

• Which AI providers will be available at launch versus added later?
• Will Apple charge AI companies for platform access or take a revenue share?
• How will Apple handle model updates and version changes from third-party providers?
• Will the feature be available globally or limited to certain markets initially?
• Could Apple eventually allow users to bring their own API keys for custom model access?

The answers to these questions will determine whether Apple's multi-model approach truly empowers user choice or simply creates a curated marketplace with Apple as gatekeeper. Given Apple's track record, expect a controlled rollout that prioritizes user experience and privacy over maximum flexibility.

One thing is clear: the era of exclusive AI partnerships in consumer technology may be ending. As AI models become increasingly commoditized and users grow more sophisticated in their preferences, platform makers like Apple must adapt. By opening Apple Intelligence to multiple providers, Apple is betting that the best AI experience isn't about picking the right model — it's about letting users make that choice themselves.

This shift could set the template for how AI integration works across all consumer platforms in the years ahead, making the fall 2025 launch one of the most consequential moments in the ongoing AI revolution.