Apple is doing something very un-Apple. According to a Bloomberg report, the company is seriously considering ditching its own large language model, the thing that powers Siri’s brain, and replacing it with Claude from Anthropic or ChatGPT from OpenAI.
Yes, you read that right. Siri, Apple’s famously underwhelming assistant, might finally get smart. But to do that, Apple may need to borrow the brains from companies that beat it to the AI finish line.
So, what’s going on?
The Reality Check: Apple’s AI Isn’t There Yet
Apple launched its long-awaited Apple Intelligence suite in iOS 18 and macOS Sequoia. That included an upgraded Siri, at least in theory. The plan was to inject Siri with actual intelligence, using a blend of on-device and server-based AI models. But internally, it looks like Apple’s own LLM just isn’t good enough.
The company has reportedly delayed the full rollout of this new Siri until sometime in 2026. And now, they’re running tests with Claude and ChatGPT to see if someone else’s model can do a better job. Spoiler: it probably can.
Who’s Winning the Test Drive?
Between the two options, Anthropic’s Claude is apparently leading the pack. Apple likes its tone, safety, and quality. OpenAI’s ChatGPT is still on the table. They already partnered up for some features in iOS 18. But Claude seems to be impressing more in private demos.
There’s even talk of Apple testing these models on its Private Cloud Compute infrastructure. That’s basically a fancy way of saying, “We’ll let them into the walled garden, but we still want control.”
Why This Is a Big Shift
Apple has spent the last two decades building a reputation around vertical integration. It controls the hardware, software, chips, ecosystem, everything. That’s why the iPhone is so smooth, why the Mac just works, and why Siri has been… well, bad.
Now Apple is flirting with a different strategy. One that looks a lot more like, “Let’s just license something that actually works.”
That’s not small. That’s Apple saying it can’t afford to be behind on AI. Not when Google’s Gemini is powering the Pixel. Not when Microsoft has OpenAI baked into Windows. And definitely not when Samsung is already licensing smarter assistants.
Privacy vs Performance: The Classic Apple Dilemma
Here’s where it gets tricky. Apple’s entire brand is built on privacy. Using third-party models means giving up some control, even if everything runs on Apple’s own infrastructure.
That’s a tightrope. Consumers love Siri for being local and private, even if it’s kind of dumb. But if Siri becomes smarter by outsourcing its brain, will Apple users still trust it?
Apple is trying to thread the needle. Keep things private. Make things smarter. And not admit that it’s playing catch-up in the biggest tech shift since the iPhone.
What’s Next?
Right now, Apple is still testing. No final decision has been made. But if they do go this route, it could fast-track Siri’s evolution. We’re talking real upgrades, real intelligence, maybe even real usefulness — before the end of 2025.
At the same time, Apple will likely continue building its own models in the background. Because long term, they’ll want to bring everything back in house. That’s just who they are.
Final Thought
This isn’t just a Siri story. It’s a turning point for Apple. For the first time in a long time, the company is willing to admit, even if not publicly, that someone else might do a better job at something core to its ecosystem.
And for us as users? It means we might finally get the Siri we were promised back in 2011. One that actually helps instead of frustrating.
About time.
Source: Bloomberg
