Tarjama Launches Pronoia V2, Leading Arabic AI Model

Tarjama has introduced Pronoia V2, a state-of-the-art Large Language Model (LLM) designed to excel in Arabic and multilingual enterprise applications, announced in a press release. Built on over 16 years of linguistic expertise, Pronoia V2 outperforms global competitors like GPT-4o, DeepSeek-V3, and Cohere in Arabic-related tasks.

Pronoia achieves a 76.8% average performance across Arabic benchmarks, significantly surpassing its competitors. It operates efficiently on a single GPU, reducing infrastructure costs while maintaining high performance. The model is available in 7B and 14B configurations, supporting text, audio, and structured data processing in both Arabic and English.

Pronoia V2 is designed for enterprise use, offering features like high-quality Arabic translations, automated contract review, and multilingual customer support. It is available immediately to enterprises via API, with options for on-cloud and on-premises deployment through Tarjama's secure AI framework. Looking ahead, Tarjama is developing Pronoia V3, aiming to further enhance its performance in Arabic AI applications.