The Technology Innovation Institute, UAE’s government research institute, has launched new versions of its large language GenAI model, Falcon 2, in collaboration with Abu Dhabi’s Advanced Technology Research Council. 

The new AI model includes two versions: Falcon 2 11B, an efficient LLM trained on 5.5 trillion tokens with 11 billion parameters, and Falcon 2 11B VLM, a vision-to-language model that converts visual inputs into textual outputs seamlessly. Both models are multilingual, but the Falcon 2 11B VLM is notably TII's first multimodal AI model, for its unique ability to perform image-to-text conversions, stated the research institute.

Interestingly, Falcon 2 11B has outperformed Meta's newly launched Llama 3, which has 8 billion parameters, and matched the performance of Google's Gemma 7B, with a score of 64.28 as compared with Gemma 7B’s 64.29 score. These metrics were independently verified by Hugging Face, which is a US-based platform that provides an evaluation tool and global leaderboard for open LLMs. Additionally, UAE’s new AI model series are open-source, offering developers worldwide unrestricted access. The research institute plans to further expand its next-generation AI models to include a variety of sizes and integrate advanced ML techniques including, 'Mixture of Experts' (MoE) to enhance their capabilities. 

AI models from the government research institute have ranked as ‘top-tier’ among powerful open-source large language models globally. They are designed to enhance market adoption in generative AI and support multiple languages including English, French, Spanish, German, Portuguese, and others. These models can interpret images and visuals, making them useful in industries like healthcare, finance, e-commerce, education, and legal sectors for tasks such as document management and digital archiving. They are also efficient, running on just one GPU, which makes them scalable and easy to deploy on devices like laptops.

You may also like:

• AI evolution in the Middle East workplaces: How prepared are talent and HR? Insights from Lars Gehrmann
• State of Cybersecurity in the age of AI: How is the Middle East preparing to deal with threats?

Commenting on the launch, H.E. Faisal Al Bannai, Secretary General of ATRC and Strategic Research and Advanced Technology Affairs Advisor to the UAE President, said, "With the release of Falcon 2 11B, we've introduced the first model in the Falcon 2 series. While Falcon 2 11B has demonstrated outstanding performance, we reaffirm our commitment to the open-source movement with it, and to the Falcon Foundation. With other multimodal models soon coming to the market in various sizes, our aim is to ensure that developers and entities that value their privacy, have access to one of the best AI models to enable their AI journey."

Dr Hakim Hacid, Executive Director and Acting Chief Researcher of the AI Cross-Center Unit at TII, added, “AI is continually evolving, and developers are recognising the myriad benefits of smaller, more efficient models. In addition to reducing computing power requirements and meeting sustainability criteria, these models offer enhanced flexibility, seamlessly integrating into edge AI infrastructure, the next emerging megatrend. Furthermore, the vision-to-language capabilities of Falcon 2 open new horizons for accessibility in AI, empowering users with transformative image to text interactions.”