Meet Saba, Mistral's AI Innovation For Malayalam, Tamil And Arabic Language Users

The development of Saba is said to be a milestone for Mistral AI, which was founded in April 2023 by three French AI researchers, Arthur Mensch, Guillaume Lample, and Timothée Lacroix.

Mistral AI Edited by
Meet Saba, Mistral's AI Innovation For Malayalam, Tamil And Arabic Language Users

Meet Saba, Mistral's AI Innovation For Malayalam, Tamil And Arabic Language Users

Mistral AI, a French artificial intelligence startup, has just launched a new AI model called Saba, specifically designed for the Middle Eastern and South Asian regions.

This groundbreaking model marks Mistral AI’s first foray into localised AI, tailored to meet the unique cultural and linguistic needs of these regions.

Saba is a small language model with 24 billion parameters, trained in foreign languages such as Arabic, Tamil, and Malayalam.

This model is designed to be lightweight and compatible with single-GPU systems, making it adaptable for a variety of use cases.

Saba can achieve speeds exceeding 150 tokens per second, making it an attractive solution for businesses and organisations looking to leverage AI in their operations.

The development of Saba is said to be a milestone for Mistral AI, which was founded in April 2023 by three French AI researchers, Arthur Mensch, Guillaume Lample, and Timothée Lacroix.

The company has gained prominence as an alternative to proprietary AI systems, with a focus on open-source innovation and democratising access to AI.

Also, read| West Asia’s Entry Into The US-China Great AI Game

The development of Artificial Intelligence (AI) in non-English languages has gained momentum among top tech players like OpenAI, Meta, and Google.

These companies are striving to maintain their lead in geography-specific applications by catering to the diverse language needs of users worldwide.

Meta has taken a significant step in this direction by introducing BOUQuET, a comprehensive dataset designed to enhance the evaluation of multilingual machine translation. Initially, BOUQuET will focus on seven non-English languages, including French, German, Hindi, and others.

Similarly, Google Deepmind, the research lab at Google, is working on Project Vaani, an ambitious initiative aimed at covering 125 Indian languages and dialects.

The goal of Project Vaani is to build an inclusive and equitable Indic AI that caters to the diverse language needs of the Indian population.

Now, Mistral AI‘s philosophy emphasises openness and innovation in the AI field. The company has gained significant traction and attention from investors, with a valuation of over $2 billion as of December 2023. In June 2024, Mistral AI secured a €600 million funding round, elevating its valuation to €5.8 billion.

The company’s partnership with Microsoft, announced in February 2024, has further expanded its presence in the AI industry.

Also, read| DeepSeek, Alibaba AI: Was Everything Planned?

Under the agreement, Mistral’s language models, including Saba, will be available on Microsoft’s Azure cloud, while the multilingual conversational assistant Le Chat will be launched in the style of ChatGPT.

The company recently announced updates for Le Chat, including the ability to create images, search for information on the internet, and introduce the Canvas system, a collaborative interface where the AI generates code and the user can modify it. Le Chat is now available on iOS and Android, making Mistral AI’s language models accessible on mobile devices.