Apple has been relatively tight-lipped about its foray into generative AI technologies, but recently, it provided a sneak peek into its ongoing development efforts in this realm. This week, it unveiled OpenELM, described as a cutting-edge open language model. These language models serve as the backbone for AI chatbots and image generators like ChatGPT, Gemini, Perplexity, and DALL-E, drawing from vast repositories of data to generate responses to user prompts, whether it's crafting emails, coding, or producing imaginative images.

While OpenELM isn't the full-fledged Apple AI product many have been anticipating, it represents a significant stride in that direction. Moreover, it offers tantalizing insights into the potential AI capabilities that might feature in the upcoming iOS 18 software for iPhones.

In a quarterly earnings call earlier this year, Apple's CEO Tim Cook hinted at the unveiling of the company's plans for generative AI sometime in 2024. Concurrently, reports surfaced suggesting that Apple redirected its focus from its longstanding electric car project to prioritize generative AI and the release of Apple Vision Pro, a wearable device that garnered attention from CNET reviewer Scott Stein for its uniqueness.

The unveiling of OpenELM adds another layer to Apple's evolving strategy. While its alignment with broader plans remains unclear, a research paper released in March sheds light on Apple's exploration of multimodal large language models capable of generating diverse content formats.

While Apple has been relatively reserved, other tech giants and startups have surged ahead in the race to develop and deploy generative AI products. Companies like Adobe, Anthropic, Google, and OpenAI are continuously pushing the boundaries, not only in understanding various queries but also in creating increasingly realistic images and videos. They are also showcasing research projects in areas like video games and music composition to entice users and secure their loyalty to their AI offerings.

OpenELM's release signifies Apple's commitment to transparency and reproducibility in large language models. The initiative aims to foster open research, ensuring trustworthy results and facilitating investigations into data and model biases and potential risks.

Apple's models, with parameters ranging from 270 million to 3 billion, demonstrate the company's investment in scalable AI solutions. While these numbers may seem modest compared to industry benchmarks like ChatGPT 3.5 with 175 billion parameters, they reflect Apple's strategic approach to AI development.

Reports hint at Apple's internal projects, including an AI chatbot named Apple GPT and a large language model called Ajax. However, the company has remained tight-lipped about these endeavours, adhering to its tradition of showcasing work once it's ready for public disclosure.

Cook emphasized Apple's methodical approach during the earnings call, highlighting the company's commitment to delivering impactful innovations without prematurely announcing them.