Apple's LLM has four variants, and even the smallest variant has performance similar to OpenAI's ChatGPT. This information comes from Apple's research, shared via Windows Central.

ReALM is an LLM developed by Apple to enhance Siri's capabilities, providing better contextual understanding. Notably, the model can process information "on-screen," meaning content currently displayed on the screen. This offers a significant advantage over ChatGPT. OpenAI's AI can process image files and PDFs but cannot read the entire screen.

realm 1712071268553222101487.jpg
Apple is the only tech giant that hasn't revealed its internal AI efforts to the public.

Apple is expected to launch four variants, including: ReALM-80M, ReALM-250M, ReALM-1B, and finally ReALM-3B. The letters M/B after the model number represent the reference number (millions and billions).

Researchers conducted performance tests on ReALM to compare it with OpenAI's GPT-3.5 and GPT-4. The results showed that even the smallest variant, ReALM-80M, was comparable to GPT-4, while higher-end models outperformed it.

The study shows that ReALM offers significant improvements over similar systems, and even the smallest model (80M) processes on-screen information 5% better.

Apple is the only tech giant that has remained silent on the booming AI technology trend since ChatGPT emerged.

Last week, Apple set the date for its Worldwide Developers Conference (WWDC), and it's highly likely that ReALM will be on the agenda this coming June.

(According to Dtrends)

Apple has set extremely stringent requirements for its suppliers for the iPhone 16. The iPhone 16 will have the thinnest screen bezels ever, but these demanding requirements could affect the launch plans for the new iPhone and increase the product's price.