Inception42 Launches Seraj, a New Arabic AI Model Built With Microsoft

Abu Dhabi based Inception42 has launched Seraj, a new enterprise AI model built jointly with Microsoft to strengthen Arabic language reasoning, dialect comprehension and cultural context inside frontier AI systems. The model is now available through Compass, Core42's sovereign AI platform, and is named after the Arabic word for light, reflecting its aim to address gaps that have long limited enterprise AI adoption in Arabic speaking markets.

Built on GPT 4.1, Seraj takes a different approach from Arabic first models built entirely from the ground up. Inception42 applied targeted mid training using curated Arabic datasets covering linguistics, cultural knowledge, safety scenarios and domain specific enterprise content, aiming to lift Arabic performance without weakening the model's existing reasoning or English language capability.

Why Enterprise Arabic AI Has Lagged Behind English

The launch addresses a problem that has quietly constrained AI deployment across the region for years. Frontier models trained primarily on English data have historically shown real gaps in Arabic linguistic precision, dialect handling and safety judgement, gaps that matter significantly in sectors such as government services, legal analysis and Islamic studies, where a single mistranslation or missed cultural cue can carry real consequences. Seraj is designed to support document understanding, summarisation, translation, question answering, workflow automation, bilingual Arabic English applications and retrieval augmented generation across government, education, legal services, media and financial services.

Ashish Koshy, CEO of Inception42, described Seraj as changing the choice organisations across the region have long faced between global AI capability and genuine Arabic performance. Rima Semaan, Director of AI and Enterprise Solutions at Microsoft UAE, said the collaboration reflects a shared commitment to making AI relevant and accessible for Arabic speaking organisations, tying the launch to the UAE's wider push to accelerate responsible AI adoption across government and enterprise.

Part of a Broader Regional Push for Sovereign, Language Aware AI

Seraj lands at a moment when Gulf governments and enterprises are increasingly weighing AI sovereignty alongside raw model capability, a theme that echoes the recent debate around the US lifting export controls on Anthropic's Fable and Mythos models, where questions of who controls access to frontier AI became central to enterprise procurement decisions. It follows Inception42's earlier work on the open source Jais Arabic language model, and reflects a broader regional bet that trust and cultural precision, not just raw benchmark performance, will determine which AI systems actually get deployed at scale inside government ministries and regulated industries across Saudi Arabia's national AI push as much as the UAE's.

What This Means for Enterprise Buyers

For CIOs and government technology leads across the GCC evaluating AI vendors, Seraj adds a credible option built specifically around Arabic performance without sacrificing English capability. As sovereign platforms like Compass mature, buyers are likely to weigh language and cultural fluency more heavily in procurement decisions, particularly in sectors such as legal services and public administration where precision in Arabic is not optional.