All of the major tech companies, with the exception of Apple, have made their generative AI models available for use in commercial settings. The business is, nevertheless, actively engaged in that area. Wednesday saw the release of Open-source Efficient Language Models (OpenELM), a collection of four incredibly compact language models—the Hugging Face model library—by its researchers. According to the company, OpenELM works incredibly well for text-related tasks like composing emails. The models are now ready for development and the company has maintained them as open source.
In comparison to models from other tech giants like Microsoft and Google, the model is extremely small, as previously mentioned. 270 million, 450 million, 1.1 billion, and 3 billion parameters are present in Apple’s latest models. On the other hand, Google’s Gemma model has 2 billion parameters, whereas Microsoft’s Phi-3 model has 3.8 billion. Minimal versions are compatible with phones and laptops and require less power to operate.
Apple CEO Tim Cook made a hint in February about the impending release of generative AI features on Apple products. He said that Apple has been working on this project for a long time. About the details of the AI features, there is, however, no more information available.
Apple, meanwhile, has declared that it will hold a press conference to introduce a few new items this month. Media invites to the “special Apple Event” on May 7 at 7 AM PT (7:30 PM IST) have already begun to arrive from the company. The invite’s image, which shows an Apple Pencil, suggests that the event will primarily focus on iPads.
It seems that Apple will host the event entirely online, following in the footsteps of October’s “Scary Fast” event. It is implied in every invitation that Apple has sent out that viewers will be able to watch the event online. Invitations for a live event have not yet been distributed.
Apple has released other AI models before this one. The business previously released the MGIE image editing model, which enables users to edit photos using prompts.