While AI might not be the focus of WWDC this year, Apple is set to make several important Apple Intelligence announcements come Monday. Here’s what’s expected, as reported by Bloomberg’s Mark Gurman.
Apple has built new, more powerful versions of its foundation models, the large language models that power Apple Intelligence features like Writing Tools, Genmoji, and text summarization.
As Gurman had previously noted:
“Versions with 3 billion, 7 billion, 33 billion and 150 billion parameters are now in active use. The 150 billion parameter model, which relies on the cloud like OpenAI and Google, is far more powerful than Apple’s on-device Tech and capable of more nuanced reasoning.”
Apple will let developers access its in-house foundation models for the first time. That means third-party apps will be able to offer features such as generating summaries, writing suggestions, and offering other generative tools using the same tech behind Apple’s Writing Tools and Genmoji.
However, developers will only have access to on-device models. Cloud models will, for now, be off-limits.
The Translate app is expanding in a big way. Apple is integrating translation into text messages and phone calls, and has also been working on bringing it to live AirPods conversations.
This is poised to be one of Apple Intelligence’s most practical consumer-facing features for the year, and it might finally put Apple on a stronger footing next to Google’s existing tools in Android.
Genmoji, Apple’s AI emoji generator, is getting a fun little upgrade. People will now be able to combine two standard emoji to generate an entirely new one.
The Shortcuts app is getting a smarter, more powerful engine. With Apple Intelligence integration, people will be able to create automations more naturally, with the system understanding more contextual prompts and turning them into actual shortcuts.
Apple is set to announce a machine learning-based battery saving mode, designed to adapt to your usage patterns in real time. It might come later in the year, as it was developed for the upcoming iPhone 17 Air (which will reportedly have a smaller battery and might need all the energy efficiency help it can get).
Last year, Apple previewed Swift Assist, an AI coding assistant built into Xcode that ran on Apple’s own AI models. However, it never launched because internal testing revealed hallucinations and reliability problems.
This time around, Apple is trying something different: a new version of Xcode that works with third-party LLMs, like Anthropic’s Claude. These models can run locally or in the cloud, bringing Xcode on par with other LLM-integrated IDEs.
Zero surprise given recent leaks and developments, but there won’t be Siri news at this year’s keynote. A revamped, LLM-powered Siri, nicknamed LLM Siri internally, is still at least a year out. When it ships, it will enable much deeper control of apps and access to personal context, but it won’t be demoed next week.
Apple has been in talks with Google to integrate its Gemini LLMs as an alternative to OpenAI’s ChatGPT, particularly for Siri and Writing Tools. However, the partnership is still pending, likely tied to the outcome of the U.S. DOJ’s antitrust ruling on the Google-Apple search deal. So, despite Alphabet CEO Sundar Pichai’s recent remarks, no Gemini announcement is expected at WWDC. Any reveal would likely come later this year.
Two Apple Intelligence efforts originally planned for this year won’t be making an appearance:
Apple has been working on a fully rebuilt Calendar app with AI scheduling features and on a previously reported new AI doctor-style Health assistant, code-named Mulberry. Due to project delays, neither is poised to be demoed next week. Both are now targeted for iOS 27 and macOS 27 in 2026, although Gurman says, ”There will be smaller changes this year.”
FTC: We use income earning auto affiliate links. More.
Check out 9to5Mac on YouTube for more Apple news:
Marcus Mendes is a Brazilian tech podcaster and journalist who has been closely following Apple since the mid-2000s.
He began covering Apple news in Brazilian media in 2012 and later broadened his focus to the wider tech industry, hosting a daily podcast for seven years.