The United Arab Emirates has launched a cutting-edge Arabic language artificial intelligence model, Falcon Arabic, as part of its growing ambition to become a global leader in AI innovation.
Developed by Abu Dhabi’s Advanced Technology Research Council (ATRC), the model is designed to reflect the full richness and diversity of the Arabic language, using a high-quality native dataset.
Falcon Arabic is said to match the performance of models up to ten times larger, representing a major breakthrough in Arabic natural language processing (NLP). In addition, the ATRC unveiled Falcon H1, a high-performance system claimed to surpass rival models from international tech giants like Meta and Alibaba, while operating with lower computational demands and requiring less technical expertise.
The launch comes as Gulf countries ramp up their AI investments. The UAE, in particular, is leveraging its close ties with the United States to advance its AI capabilities. This follows a recent visit by U.S. President Donald Trump to the region, during which an AI agreement was announced granting the UAE access to advanced American AI semiconductors, an important step towards building a robust AI infrastructure.
In addition to Falcon Arabic, Abu Dhabi-based AI firm G42, through its subsidiary Inception, has also made significant strides with the release of Jais, an open-source Arabic-English bilingual large language model. Named after Jebel Jais, the UAE’s highest peak, the 13-billion-parameter model was developed in collaboration with the Mohamed bin Zayed University of Artificial Intelligence (MBZUAI) and Cerebras Systems. Trained on a 395-billion-token dataset, Jais is designed to power a wide range of NLP tasks such as machine translation, text generation, and question-answering.
Also Read: Brazil, Chile to Develop AI Model Reflecting South American Identity
G42 has further expanded its AI portfolio with JAIS 70B, a 70-billion-parameter model, and a suite of 20 additional models tailored for Arabic and multilingual applications. These models, ranging from 590 million to 70 billion parameters, were trained on up to 1.6 trillion tokens covering Arabic, English, and code, and are set to support generative AI services across diverse sectors, including customer service, content creation, and data analytics.
These developments underscore the UAE’s strategic focus on Arabic-language AI, positioning the country at the forefront of technological advancement in the region.
