mbzuai

LLMVoX: Autoregressive Streaming Text-to-Speech Model for Any LLM

A reproducible, source-grounded admissions eligibility engine that models program-specific requirements from official university sources, validates applicant inputs, and produces explainable eligibility decisions.

[NAACL 2025 🔥] CAMEL-Bench is an Arabic benchmark for evaluating multimodal models across eight domains with 29,000 questions.

Revision site for the MBZUAI BSc screening exam: lessons, 1,300+ questions with mark schemes, endless generated practice and timed mocks.

Bahar is a chatbot web application powered by JAIS, an Arabic-centric Large Language Model developed by MBZUAI and Inception AI. This project, developed during the Emirati AI Boot Camp 2024, showcases the integration of LLMs into real-world applications, featuring a user-friendly interface and the ability to generate text in both Arabic & English.

LLMVoX: Autoregressive Streaming Text-to-Speech Model for Any LLM