Falcon-Emirati-7B focuses on Emirati Arabic
TII’s seven-billion-parameter model aims to deliver responses in the Emirati dialect and cultural context, as well as answering correctly.

AT A GLANCE
Key points
- Based on the seven-billion-parameter Falcon-H1-Arabic
- Producer reports 84.83% accuracy on Alyah
- Licensing and distribution terms are unclear in the announcement
What is changing?
The Technology Innovation Institute (TII) introduced Falcon-Emirati-7B on October 6. Based on the seven-billion-parameter version of Falcon-H1-Arabic, it was adapted using Emirati websites and forums, Modern Standard Arabic cultural sources and controlled synthetic dialect examples. The team says it used native-speaking evaluators to assess naturalness and cultural appropriateness.
The model achieved 84.83% accuracy on TII’s 1,173-question Alyah test. In a comparison of open-ended dialect generation judged by Gemini 3.7 Flash, it reported a score of 0.52. These are the producer’s own tests, not an independent benchmark.
The announcement provides a trial link, but does not clearly specify licensing or weight distribution terms. The model may be promising for local customer service, internal assistants and Arabic RAG systems, but access terms and real-world accuracy need checking before production use.
Explore the original sources
Source date: 6 October 2026.
Published by AI Radar: 6 Oct 2026, 13:06 (Istanbul, UTC+3). Updated: 6 Oct 2026, 18:44.
FOLLOW THE TOPIC
Topics and tags
Share your discovery.
Send this story to a friend.
Reader comments
What do you think about this development? Share your experience or question. Comments are published after editorial approval.
Your name/nickname and comment become public if approved. Read: Privacy notice.






Loading comments…