Cohere Labs have introduced Tiny Aya, a brand new line of open, weight multilingual AI models. These models help in more than 70 languages and can work very efficiently even on limited everyday hardware such as laptops or other edge devices. They do not need an internet connection, thus, it is a big step towards making AI accessible to everyone and working offline in situations where the internet connection is limited. The Tiny Aya collection consists of a 3. 35 billion, parameter base model together with four specialized derivatives: TinyAya, Global for broad and versatile instruction following, TinyAya, Earth for African languages, TinyAya, Fire for South Asian languages, and TinyAya, Water for Asia Pacific, West Asia, and European languages. This way the models offer a wide language coverage and also include region, specific linguistic nuances while remaining very small in size. This lowers the computational barriers for developers and researchers. “This approach allows each model to develop stronger linguistic grounding and cultural nuance, creating systems that feel more natural and reliable for the communities they are meant to serve. At the same time, all Tiny Aya models retain broad multilingual coverage, making them flexible starting points for further adaptation and research,” the company said in a statement.
Also Read: OpenAI Unveils GPT-5.2, Elevating AI for Professional Knowledge Work
The Tiny Aya models were announced at the India AI Summit 2026 and trained on a modest cluster of 64 Nvidia H100 GPUs, reflecting Cohere’s goal of balancing performance with practical deployment needs, particularly in linguistically diverse regions like India where consistent connectivity can be a challenge. These models support a range of major South Asian languages including Bengali, Hindi, Punjabi, Urdu, Gujarati, Tamil, Telugu and Marathi and can be downloaded from platforms such as HuggingFace, Kaggle, Ollama, and the Cohere Platform, empowering developers and local communities to build offline translation tools, native language applications, and educational tools without cloud dependencies. The open-weight nature of Tiny Aya aligns with Cohere’s broader mission to expand multilingual AI access and foster community collaboration, lowering technical barriers while promoting AI that better reflects cultural and linguistic diversity worldwide.
