The Rise of Gemini: How Googles All-in-One AI Leaves Siri Behind

Google is developing Gemini as a unified artificial intelligence platform with a clear architecture and business model, while Apple is struggling with Siri, caught between technical debt, internal upheaval, and the weight of its own legacy.

At the Google I/O 2025 event, Gemini took center stage. The Gemini brand appeared in nearly every demonstration, on all devices, and in every statement made by Sundar Pichai and his team. According to Google, seven million developers are currently working with Gemini models, and the monthly token volume surged from 9.7 trillion to 480 trillion in just a year—a fiftyfold increase.

Apple’s story with artificial intelligence tells a different tale. Siri continues to operate on a hybrid architecture that insiders refer to as having «two brains»: an old, rule-based system augmented by a limited generative language model. This combination of outdated code and new technologies has resulted in what developers describe as an endless «whack-a-mole» scenario with bugs: fix one, and three more pop up elsewhere, as noted by Bloomberg.

However, the core issue runs deeper than it might appear. The very architecture hinders any genuine progress. Apple is quietly developing an entirely new LLM architecture in Zurich that will eventually replace the old foundation of Siri, but until then, Siri remains in limbo.

As a result, the long-awaited Siri 2.0, initially set to launch with iOS 18.5, has been postponed to iOS 19, which won’t be released until at least spring 2026. Internal testing indicates that Siri’s key functions only work 66%-80% of the time. Reuters reports that Apple CEO Tim Cook replaced former AI chief John Giannandrea with Mike Rockwell, a hardware manager known for his work on the Vision Pro.

The contrast between these two systems is stark. Siri was introduced in 2011 as a feature to help sell iPhones. By 2025, Gemini has evolved into a cloud product designed to generate revenue independently. Google offers two premium tiers for Gemini: Gemini AI Pro at $19.99 per month and AI Ultra at $249 per month—both directly linked to the latest models. Millions of developers also access Gemini via its API.

Google’s path to Gemini has been paved over several years: starting with LaMDA in 2021, followed by the multimodal MuM, the experimental Bard in 2023, and Gemini 1.0 released at the end of that same year. With Gemini 1.5 Pro, Google finally caught up with OpenAI. Now, with 2.0 Pro and 2.5 Pro, featuring the new Deep Think mode, Google confidently leads the field.

Moreover, Google is rolling out new services like Flow (an AI-powered video creation tool) and Jules (an asynchronous coding agent), built on Gemini, Veo, and Imagen. Developers gain instant access to these tools through a comprehensive SDK and cloud access via Vertex AI. In contrast, Siri remains a closed system with a limited intents API and basic third-party app integration.

Gemini is not just another version of Siri; it is fundamentally different: a comprehensive, relatively open, and actively evolving platform. While Apple continues to grapple with its 14-year-old voice system, Google is already deploying multimodal agents integrated into camera applications, chat interfaces, programming environments, and XR headsets. AI has transitioned from being a feature to becoming the foundation.

[Source](https://the-decoder.com/google-gemini-is-everything-siri-never-was/)