Abu Dhabi: The United Arab Emirates has launched Jais 3, a new family of artificial intelligence models designed to understand and generate Arabic across different regional dialects. The technology was introduced on October 15, 2026, by the Mohamed bin Zayed University of Artificial Intelligence (MBZUAI) and Inception42, in collaboration with Cerebras Systems.
The new AI family includes six open language models, ranging from 900 million to 375 billion parameters, along with a separate speech-recognition system called Jais Speech. The technology is designed to help users write content, transcribe conversations, prepare documents and presentations, and carry out tasks through connected digital tools.
One of its major features is Jais Speech’s ability to recognise Arabic dialects spoken across 19 Arab countries. It can also process conversations where speakers switch between Arabic, English and French, making the system more useful for everyday communication.
According to its developers, Jais 3 has been designed for Arabic-speaking individuals, businesses and government organisations. The models have been released through Hugging Face for developers and researchers, while claims about their performance compared with other AI systems are based on evaluations published by the developers.
