Cohere Launches New Multilingual Models at India AI Summit
At the India AI Summit, Cohere, an Enterprise AI company, unveiled a new series of multilingual models called Tiny Aya. These models, developed by Cohere Labs, are open-weight, making their code accessible for public use and modification. They support over 70 languages and are designed to run on everyday devices like laptops without the need for an internet connection.
Expanding Language Support
The Tiny Aya models cater to South Asian languages such as Bengali, Hindi, Punjabi, Urdu, Gujarati, Tamil, Telugu, and Marathi. The base model boasts 3.35 billion parameters, showcasing its size and complexity. Additionally, Cohere introduced specialized versions like TinyAya-Global for broader language support, TinyAya-Earth for African languages, TinyAya-Fire for South Asian languages, and TinyAya-Water for Asia Pacific, West Asia, and Europe.
Enhanced Linguistic Grounding
The company emphasized the importance of linguistic grounding and cultural nuance in their models, aiming to create natural and reliable systems tailored to specific communities. Despite their focus on regional languages, the Tiny Aya models maintain broad multilingual coverage, facilitating further adaptation and research.
Cohere highlighted the efficiency of these models, trained on a cluster of 64 H100 GPUs, for developers and researchers targeting native language-speaking audiences. The models can operate offline, enabling offline translation capabilities and requiring minimal computing power compared to similar models.
Applications in Diverse Regions
Offline-friendly features of the Tiny Aya models offer diverse applications in linguistically diverse countries like India, eliminating the need for constant internet connectivity. These models are available on platforms like HuggingFace, Kaggle, and Ollama for local deployment, with training and evaluation datasets released on HuggingFace for further exploration.
Future Plans
Cohere’s CEO, Aidan Gomez, hinted at the company’s plans to go public soon, following a successful year with $240 million in annual recurring revenue and consistent growth throughout 2025. As Cohere continues to innovate in the AI space, their multilingual models are set to revolutionize language processing and communication in diverse communities worldwide.