The Birth of “Thawra”
This model was launched by the Syrian brothers Hani and Saeed Shehabi, who left Syria before 2011 and then studied technology in Germany before moving to work in major technology companies. After years of experience, they decided to develop an Arabic model that gives top priority to privacy, offers high accuracy, and is environmentally friendly.
A Radically Different Technology
“Thawra” operates using an open-source model from the “Mixture of Experts” category, GLM-4.5 Air, and includes more than 100 billion parameters. But interestingly, it only activates about 12 billion with each query, which distinguishes it from traditional models that use their full computational power in every interaction.
According to its official website, this design makes it faster and more energy-efficient by up to 93% compared to “ChatGPT,” in addition to being less expensive to operate.
The model also relies on the “OpenAI” developer package, making it compatible with most programming interfaces used in modern applications.
Privacy Comes First
“Thawra” operates on fully secured DigitalOcean and TogetherAI servers, making it beyond the control of giant corporations or governments. The company emphasizes that all conversations are protected, with a limited portion retained for technical purposes.
Arab Identity and a Clear Message
Hani Shehabi says in statements to the “Inside Brka” website that their goal was to make a real impact in the technology sector, especially in light of what he described as the “Western blackout” on issues in the region, especially Palestine. He adds that “Thawra” came as a reaction to the growing concern about the absence of an ethical framework in some popular artificial intelligence models.
