More than 400 million Arabic speakers and roughly 2 billion Muslims worldwide have historically been underserved by mainstream generative AI, built primarily around English-language data and Western cultural context. Humain IQ, the portfolio behind HUMAIN’s flagship AI model products, set out to close that gap directly, and its first release has already been independently verified as a genuine milestone rather than just another regional chatbot.
The Launch That Started It All
On 25 August 2025 in Riyadh, HUMAIN, a Public Investment Fund company, launched HUMAIN Chat, a next-generation Arabic conversational AI app powered by ALLAM 34B, the company’s flagship Arabic large language model. Available on web, iOS, and Android, the app launched first in Saudi Arabia as a national milestone, with regional rollout across the Middle East and global expansion planned to follow.
Why ALLAM 34B Is Being Called a Milestone, Not Just a Model
ALLAM 34B builds on earlier models developed by the National Center for AI under the Saudi Data & AI Authority’s national mandate, and this latest version has been independently verified by Cohere on the MMLU benchmark as the world’s most advanced Arabic large language model built in the Arab world. It is Arabic-first by design while remaining fully bilingual in English, trained on one of the largest Arabic datasets ever assembled, then refined with input from more than 600 domain experts and 250 evaluators.
The Team Behind the Model
The scale of human expertise behind ALLAM 34B is itself notable. Over 120 AI specialists, including 35 PhD holders, with a 50/50 gender ratio, built the model from the ground up, hosted in Saudi Arabia by Saudi talent working alongside global specialists.
Features Built for Real-World Arabic Use
- Real-time web search for always up-to-date knowledge
- Arabic speech input supporting multiple regional dialects
- Seamless bilingual switching between Arabic and English within the same conversation
- Conversation sharing to enable collaboration and reuse
- Full compliance with Saudi PDPL, hosted end-to-end on HUMAIN’s own infrastructure within the Kingdom
Why This Matters for Digital Inclusion
HUMAIN Chat is designed to close what the company describes as a historic gap in digital inclusion, enabling linguistic equality for a population long underserved by generative AI tools built around other languages and cultural assumptions. HUMAIN CEO Tareq Amin described the launch as “a point of pride for Saudi Arabia,” emphasizing that the technology proves globally competitive AI can be rooted in the Kingdom’s own language, infrastructure, and values, built by Saudi talent from the ground up.
The Broader Vision for Arabic AI
This launch represents only the first release within the broader Humain IQ portfolio, a new generation of AI model products and solutions the company describes as combining scientific depth with responsible design. As adoption grows across the region, the underlying model is expected to keep improving through consumer engagement, following the same pattern seen with other large language models that grow more capable the more they are used.
A Model Rooted in Cultural and Religious Context
What distinguishes this effort from a simple translation layer over an existing model is its grounding in Islamic, Middle Eastern, and cultural nuance, reflecting genuine Arabic AI development rather than an adapted Western system. This distinction matters for anything from religious and cultural queries to everyday conversational nuance that a generic model would likely miss entirely.
A Call to Action for the Region
HUMAIN has framed the Saudi launch of HUMAIN Chat as more than a product release, describing it as a collective effort inviting every citizen to use, test, and help shape the app into the region’s leading Arabic AI tool. Since large language models generally become smarter and more capable the more they are used, this early domestic engagement is expected to directly influence how quickly the model improves ahead of its planned regional and global rollout.
In the End
Building the world’s most advanced Arabic-first AI model required more than translation, it required a dedicated team, a massive Arabic dataset, and genuine cultural grounding from the start. Have you tried an AI tool built specifically around your own language and cultural context yet?
