26 January, according to Bloomberg, by Google Gemini A new generation driven by technology depth Siri The iOS 26.4 test version will be displayed as soon as next month and will be recreated for several core applications。
Mark Goulman, a journalist for Bloomberg, said yesterday in his column in Power On, that he is not the only one who is not the only one in the worldappleContacts with several model suppliers, including Anthropic and OpenAI, which were overpriced, were initiated in mid-2025, while the latter were in strategic conflict with active apple-drilling engineers and hardware configurations。
Eventually, Apple chose Gemini, in part because the United States court ruled last September that there was no need to split the search partnership between Apple and Google to clear the way for further cooperation between the parties。
On a specific point of time, Gülman believes that the first Siri functions, supported by Gemini, will follow iOS 26.4 next month into the testing phase and are scheduled to be officially delivered between March and April this year。
This version of Siri will run on the Apple's Pilot Cloud Computer server, code-named Apple Foundation Models version 10, which is approximately 1.2 trillion parameters。
This year, WWDC will be seen in a much bigger upgrade. Apple is developing a new Siri architecture, code-named "Campos", to be launched in iOS 27, iPadOS 27 and macOS 27, with a stronger contextal understanding, continuous dialogue and deep integration into core applications such as Safari, TV, Health, Music, podcasts。
At the same time, the inside of Apple's AI organization is undergoing major changes. With the departure of the former Senior Vice President of Machine Learning and Artificial Intelligence Strategy, John Giahan Dreyya, the head of software engineering, Craig Federridge, took over the direction of AI and promoted cooperation with Google。
Some of the original projects, such as the Global Knowledge Questions and Answers and the AI version of the Safari Upgrading Programme, based on internal models, have been scaled down or suspended, but may be restarted before WWDC。
The report also mentioned that apples are talking about getting a future version of Siri to run directly on Google clouded TPU to enhance performance and speed of response. At the same time, apples are still developing higher performance self-study servers to support long-term cloud-based AI layouts。
Apple has had relatively limited new AI functionality since it was launched last year, with only a small number of up-to-up updates in Apple Music and Apple Watch applications。
As internal model development has been blocked, brain drain has intensified, and Siri’s long-delayed upgrading programme has been delayed, Apple re-evaluated its AI route in the second half of last year and finally decided to work with Google to introduce Gemini into the bottom structure of Siri and Apple Industries。
As a new generation of Siri is coming to light, apples are trying to close the gap with products like ChatGPT and Gemini in the generation AI competition。