Rwanda’s Tunga AI Project Releases Open-Source Kinyarwanda AI Models and Data

Kigali, Rwanda. A major milestone in Rwanda’s digital transformation was announced this week as the Tunga Agricultural Chatbot initiative released a suite of open-source AI models, datasets, and code tailored for the Kinyarwanda language.

The effort is a collaborative project involving the Digital Transformation Center Rwanda, Rwanda Centre for the Fourth Industrial Revolution, KiNLP, and the Ministry of Agriculture & Animal Resources.

What Was Released

The open-source resources enable developers, researchers, and innovators to build AI-powered tools that understand and generate Kinyarwanda language content. The release includes:

• Text-to-Speech Model: A high-quality synthetic voice model for Kinyarwanda, enhancing accessibility and interaction.
• KinyaCOLBERT Free: The first free Kinyarwanda embedding and information retrieval model, critical for building advanced chatbots and search tools.
• Voice Dataset: A collection of Kinyarwanda audio for training both Text-to-Speech (TTS) and Speech-to-Text (STT) systems.
• Information Retrieval Dataset: Used to train the KinyaCOLBERT model, improving retrieval accuracy.
• KinyaBERT: A foundational BERT-style language model designed for natural language processing tasks in Kinyarwanda.
• Source Code: All relevant code to support the models and datasets listed above, enabling developers to adapt and extend the tools.

These resources are freely accessible to the public, encouraging broader use and development of AI solutions that speak Kinyarwanda.

A Milestone for Local AI Innovation

The open-source release marks a significant step toward inclusive and context-relevant artificial intelligence in Rwanda. By making language models and training data available in Kinyarwanda, a language spoken by over 13 million people, the initiative paves the way for applications that are more aligned with local needs.

Developers can now build tools such as:

• Local language chatbots
• Voice assistants for agricultural extension services
• Educational tools with Kinyarwanda TTS
• AI-powered search and retrieval systems

This work reflects a shift toward practical digital transformation, not just adopting global technologies but adapting them for local contexts.

Why It Matters for Rwanda and Beyond

Open-source AI for African languages is still rare. By releasing these models and datasets publicly, Rwanda positions itself as a leader in African AI innovation, helping close the language gap in technology.

These tools also support the broader goals of:

• Digital inclusion
• AI literacy and capacity building
• Agricultural productivity through intelligent tools

Developers, startups, and researchers can begin using these resources immediately to build products that are relevant, accessible, and locally meaningful.