Apple confirmed this week that the version of ChatGPT that powers parts of Apple Intelligence will leap from OpenAI’s GPT-4o to GPT-5 — but only once iOS 26, iPadOS 26 and macOS Tahoe 26 ship to users. For anyone who’s been refreshing the OpenAI blog and Apple news pages like it’s a streaming queue, that means you’ll probably see GPT-5 in your iPhone, iPad or Mac in the next major Apple software update — not the minute OpenAI flipped the GPT-5 switch.

OpenAI rolled GPT-5 out to ChatGPT users this week, making it the default model inside ChatGPT and available even to free users. The company has been clear that GPT-5 is a big step for the product line: faster, more capable at reasoning and coding, and positioned as the “most useful” model yet.

But Apple told reporters that the ChatGPT hookup inside Apple Intelligence won’t start tapping GPT-5 until the next wave of operating-system updates: iOS 26, iPadOS 26 and macOS Tahoe 26. Those updates are expected in the fall, so there’s a short lag between OpenAI’s public release and Apple’s enterprise of rolling out the model inside its assistant features.

That lag raises a question lots of readers will have: will GPT-5 be available in Apple’s developer and public betas? Apple has not publicly confirmed whether the model will show up in early betas or be held for final releases; the company’s comment to reporters was focused on the software milestone rather than the beta timeline. In practice, public testing windows will probably show us the answer within a few weeks.

Apple Intelligence is already more than a marketing label — it’s a bundle of features that take Apple’s device data, local processing and third-party models and combine them into things like smarter Mail summaries, on-device actions for Siri, and Visual Intelligence, a Google Lens-style feature that recognizes and annotates what’s in your photos. Right now, the ChatGPT connection inside that system relies on GPT-4o; swapping to GPT-5 should, in theory, make the handoff better at multi-step reasoning, answer synthesis and some coding and creativity tasks.

In plain terms: you might get smarter, clearer replies from Apple Intelligence when it calls ChatGPT for help (for example, when Siri needs to explain a dense email thread or summarize a webpage). Visual Intelligence could produce richer captions or more accurate interpretations when a cloud model is used. That said, Apple’s system mixes on-device processing with cloud models, so what you see will depend on both Apple’s layout of the feature and OpenAI’s model behavior.

Apple’s cadence for shipping major OS updates is driven by testing, privacy reviews, and ecosystem stability — not by which models are trending on X. Integrating a new, much more capable model like GPT-5 requires engineering work (API changes, latency testing), product decisions (where to route which prompt), and privacy/security reviews (what data gets sent off-device). Apple’s slower release timetable here likely reflects the company’s desire to avoid a rollout that’s faster than its ability to keep those guarantees intact. That’s Apple’s playbook: ship features when they’re ready for millions of users. (It’s also why the exact beta timing matters to apps and enterprises.)

OpenAI says ChatGPT is used by roughly 700 million people every week, a scale that makes any change to the default model consequential — not just for individual users but for how partners like Apple design product experiences. Swapping a widely used component inside Apple’s assistant ecosystem to GPT-5 is therefore a material change: it affects latency expectations, content moderation flows and the types of answers people will expect from their devices.

New models don’t automatically fix every problem. Early reactions to GPT-5 have been mixed: some testers praise its reasoning and coding chops, others note differences in “personality” or edge-case behavior — and OpenAI itself has already started to tweak how models are offered after the initial rollout. The company briefly replaced GPT-4o with GPT-5 as the default and then signalled it would keep legacy models available for users who prefer them. That suggests both OpenAI and Apple will be iterating on which model is used where, and under what conditions.

From a user perspective, the practical takeaway is simple: expect Apple Intelligence to feel a little sharper when Apple flips the GPT-5 integration on, but don’t expect an overnight revolution. There will be follow-up tweaks, and the feature will be constrained by Apple’s UI, privacy stance and the decisions it makes about when to send data to the cloud.

The GPT-5 upgrade inside Apple Intelligence is coming — and it’s a reminder that the AI arms race is now partly a software-update race. OpenAI moved first by making GPT-5 available to ChatGPT users; Apple is moving at its own tempo to bake that power into products millions use. The wait may be a short one — but the way Apple integrates GPT-5 will determine whether it feels like an incremental power boost or a genuinely different kind of assistant.