Apple uses Google's Gemini to update Siri

According to Bloomberg, Apple has completed plans to implement a customized 1.2T parameter version of Google's Gemini model for its long-delayed Siri upgrade, paying about $1 billion a year to license the technology. The specifics: Gemini, which operates on Apple's Private Cloud Compute architecture to protect user privacy, will manage summarization and multi-step planning within Siri. Additionally, Apple tested models from Anthropic and OpenAI, whose 1.2T parameter count was significantly higher than the 150B utilized in the current Apple Intelligence model. According to Bloomberg, Apple plans for Google to be a "behind-the-scenes" tech supplier, and the collaboration is "unlikely to be promoted publicly." Apple intends to employ Gemini as a temporary solution while it develops its own capable internal model, and the new Siri might be available as early as next spring. Gemini is the model that is expected to introduce the voice assistant into the AI world (at least in part) following years of delays and uncertainty surrounding Siri's upgrade. Apple sees the shift as transitory, but given the company's problems and employee turnover, developing its own answer doesn't seem likely.