In addition to leveraging Gemini to power features within its services and apps, Google makes its LLM available to outside developers. It has been reported that Apple and Google are in negotiations to license Gemini for the iPhone.
“Active negotiations to let Apple license Gemini, Google’s set of generative AI models, to power some new features coming to the iPhone software this year” are reported by Bloomberg. The company that powers Microsoft’s AI capabilities, OpenAI, has also had conversations with Apple.
As stated in today’s report, Gemini can be used for text and image generation, indicating that Apple is specifically seeking partners for cloud-based generative AI. With the impending release of “iOS 18, Apple is working to provide its own on-device AI models and capabilities.”
It’s still early in the talks, so it’s unclear what the AI will be called. The current partnership between the two companies—default search engine—would be greatly expanded by this.
Taking a look at the rest of the market, Google and Samsung announced a partnership in February so that the Galaxy S24’s voice recording and note-taking apps would have access to Gemini power summarization features. Additionally, Samsung’s photo gallery app uses Imagen 2 text-to-image diffusion for generative editing. Samsung is utilizing an on-device version of Gemini, but all of those features necessitate server-side processing.
Google provides Gemini in three sizes; the majority of first- and third-party apps use Pro. Whereas 1.0 Ultra powers the premium Gemini Advanced tier, Gemini 1.0 Pro powers the free version of gemini.google.com.
While Google previewed Gemini 1.5 in mid-February, which features a significantly larger context window that facilitates the absorption of more information, Gemini 1.0 is still available in stable. The “output more consistent, relevant, and useful” as a result may result.
- Gemini Ultra: The biggest and most powerful model for extremely complicated jobs
- Gemini Pro: The most effective model for scaling in various tasks
- Gemini Nano: The most effective model for tasks performed on a device
Although multiple providers, including OpenAI, are mentioned as possibilities in today’s report, Bloomberg predicts that a deal will not be announced until WWDC in June.