Riyadh (UNA/SPA) – A recent study confirmed the pivotal role of large Arabic language models in enabling the presence of the Arabic language in digital spaces and enhancing its ability to compete globally, which supports the presence of the Arabic language in the digital environment and accelerates the adoption of innovation in institutions, indicating that the Kingdom tops the list of countries developing Arabic language models in 2025.
The study, conducted by the Saudi Data and Artificial Intelligence Authority (SDAIA), in cooperation with the King Salman Global Academy for the Arabic Language, aimed to support the development of an artificial intelligence system in the Arabic language, and to identify the requirements for developing models that are more capable of understanding the Arabic language and its various dialects, generating content, and executing instructions.
The study addressed the history of the development of Arabic linguistic models from their beginnings based on rule-based systems before 2000 AD, through statistical models and neural networks, up to the stage of large linguistic models and their contemporary generative applications during the period from 2022 to 2025 AD, during which this stage witnessed the launch of dozens of Arabic models, including dialogic and generative models aimed at supporting the Arab need in the technical, educational and knowledge fields.
The study monitored more than (53) Arabic linguistic models up to the first quarter of 2025 AD, and the Kingdom of Saudi Arabia topped the list of countries developing these models. International bodies showed remarkable interest in developing linguistic models that support the Arabic language. The analyses showed a weakness in investment in Arabic linguistic models that support audio and visual forms, despite their future importance, as (81%) of these models were mono-media and dealt with texts only, while the percentage of multimedia models was (7%).
In terms of capabilities, the study showed that the Arabic language models included three main tasks: language comprehension, content generation, and conversation and instruction execution; while cognitive and reasoning abilities, multilingualism, and software support remain at a low level compared to global language models.
According to the results of the standard scale assessment (Balasam) issued by the King Salman Global Academy for the Arabic Language, which compares the performance of Arabic language models with their global counterparts in Arabic language tasks, the global models showed superiority in the majority of language skill categories, while at the same time the results reflected promising strengths for some Arabic models in some qualitative tasks, and slightly excelled in the summarization skill, while providing similar performance in creative writing and reading comprehension tasks.
The study reviewed the current status of Arab models, noting the existence of developed models in Arab countries, most notably Saudi Arabia and the UAE, in addition to models developed in international bodies supporting the Arabic language. It revealed gaps, most notably the limited size of the models and the number of their parameters compared to global models, the lack of comprehensive Arabic data, and the scarcity of specialized Arabic benchmarks for performance evaluation.
The study laid out a roadmap outlining practical steps to achieve leadership for large Arabic linguistic models, by focusing on providing high-quality and comprehensive Arabic data for various dialects and fields, developing multi-capacity linguistic models of diverse sizes, building Arabic benchmarks to assess the quality of models, in addition to supporting the adoption of Arabic models locally through governmental and private institutions, and disseminating them for community use.
This study comes within the framework of cooperation between “SDAIA” and the King Salman Global Academy for the Arabic Language, as a qualitative step that reflects the Kingdom’s interest in combining linguistic and cultural identity with technological development, and working to ensure the presence of the Arabic language in the global artificial intelligence system, which enhances the Kingdom’s position as a leading regional center in developing Arabic language technologies and enabling digital Arabic content.
(is over)
