عربي

Tarjama's latest edition of Pronoia outperforms global players

Arabic

Tarjama's latest edition of Pronoia outperforms global players
  • Tarjama, an Arabic language technology company based in the UAE, has launched Pronoia V2, a state-of-the-art Large Language Model (LLM) designed to deliver unmatched performance in Arabic and multilingual enterprise applications. 
  • Pronoia is an advanced large language model (LLM) specifically designed for Arabic enterprise applications that provides unparalleled accuracy and contextual understanding in sectors including legal, medical, and business.
  • Founded in 2008 by Nour Al Hassan, Tarjama has been committed to empowering Arabic-speaking markets through innovative language technology solutions.
  • Pronoia registered a 76.8% average performance across Arabic benchmarks, outpacing GPT-4o (58.05%), DeepSeek 70B (53.6%), and Cohere 7B (55.3%). 

Press release:

Tarjama, the leading language solutions technology company in the MENA region, announced today the launch of Pronoia V2, a state-of-the-art Large Language Model (LLM) designed to deliver unmatched performance in Arabic and multilingual enterprise applications. 

Built on over 16 years of linguistic expertise, Pronoia now ranks as the most accurate and cost-effective Arabic AI solution, outperforming GPT-4o, DeepSeek-V3, and Cohere on Arabic-related tasks. Pronoia sets a new benchmark for Arabic AI by combining high-performance language processing with enterprise-grade security and efficiency. 

Built to operate on just one GPU, Pronoia delivers best-in-class output while significantly reducing infrastructure costs. "Arabic has been underserved by AI for too long," says Nour Al Hassan, CEO and Founder of Tarjama. "Pronoia changes that. It's not just smarter — it's faster, more secure, and cost-effective." 

Performance That Speaks Volumes

Pronoia registered a 76.8% average performance across Arabic benchmarks, outpacing GPT-4o (58.05%), DeepSeek 70B (53.6%), and Cohere 7B (55.3%). 

It also registered an 18.75% performance gap over GPT-4o on average, a 23.2% performance advantage over DeepSeek 70B, and a 21.5% lead over Cohere 7B. 

It further achieved exceptional results on specialised benchmarks, scoring 92.4% on MT ARC Easy and 83.4% on MT Openbook QA.

Iyad Ahmad, CTO of Tarjama, explains, "Size doesn't always mean strength. We've proven that thoughtful architecture and precision training can outperform sheer scale." 

Enterprise-Ready Solutions

Pronoia enables businesses to streamline operations through high-quality Arabic translations, automated contract review, multilingual customer support, regional content generation, and instant document processing. "We've witnessed firsthand how businesses struggle with Arabic content management," notes Al Hassan. "Pronoia turns day-long processes into minute-long tasks, fundamentally changing how organisations operate in Arabic-speaking markets." 

Built for Business

Designed with enterprise needs in mind, Pronoia features enterprise-grade encryption, flexible deployment options, custom model training capabilities, and seamless platform integration. Available in both 7B and 14B configurations, Pronoia processes text, audio, and structured data in both Arabic and English, with larger versions in development to meet evolving business demands. "This isn't just a language model—it's a comprehensive business solution built from the ground up with Arabic at its core," says Al Hassan. Pronoia is available immediately to enterprises via API, with on-cloud and on-premises deployment, through Tarjama's secure AI framework. 

Looking Ahead: Pronoia Vv32

Tarjama’s team is currently developing Pronoia Vv32, aiming to raise the benchmark even further with an 80% average across Arabic evaluation datasets and a 94+ COMET score for translation quality, reinforcing its leadership in Arabic AI performance.

Thank you

Please check your email to confirm your subscription.