Uganda-based technology organisation Sunbird AI has introduced Sunflower v2, an artificial intelligence (AI) suite designed to work with 67 African languages, in an effort to make digital technology more accessible to people across the continent.
The suite was released on September 24, 2026, expanding on an earlier version that focused on 31 languages spoken in Uganda.
According to Sunbird AI's official announcement, Sunflower v2 consists of several AI models designed for different language-related tasks, including speech recognition, translation, text processing and voice generation.
AI That Can Understand and Respond to African Languages
One of the key features of Sunflower v2 is its ability to process spoken language, convert speech into text and generate spoken responses.
Its SunflowerASR model is designed to recognise speech and convert it into written text across 51 African languages.
Another component, Sunflower TTS, converts written text into speech and supports 20 languages and voices.
The suite also includes Sunflower-Gemma4-E2B, a model designed to operate on mobile devices without requiring an internet connection. It can receive spoken instructions, process them and generate voice responses.
Sunbird AI says this model has been tested in Luganda and Acholi through its Sunbird Tutor application.
Kinyarwanda Among the Supported Languages
Kinyarwanda is among the 67 African languages covered by Sunflower v2. Other languages include Swahili, Hausa, Amharic, Shona and Zulu.
This development could make digital services more accessible to people who prefer communicating in their native languages rather than relying on English or other widely used languages in technology.
Sunbird AI says its work aims to reduce the gap between languages that have large amounts of digital data available for AI training and those that remain underrepresented in artificial intelligence development.
However, the organisation acknowledges that performance varies across languages because some have limited training data. As a result, responses generated in certain languages may require additional review by fluent speakers.
More technical information is available in Sunbird AI's documentation on Sunflower v2.
Released as an Open-Source AI Suite
Sunflower v2 has been released as an open-source suite, allowing researchers, developers and organisations to access the models, adapt them and build other applications using the technology.
The models are available through platforms such as Hugging Face, while Sunbird AI also provides ways to test and integrate them into other digital services.
This approach could support the development of applications, websites and other technologies that communicate in African languages.
A Step Towards Greater Representation of African Languages in Technology
Africa is home to more than 2,000 languages, yet many remain underrepresented in modern digital technologies, particularly in artificial intelligence.
This creates challenges for people who cannot easily access digital services because those services are primarily developed in languages they may not speak or understand well.
The introduction of Sunflower v2 represents an effort to address this gap by developing AI tools that support languages used across the continent.
However, the technology's wider impact will depend on continued improvements in model performance, the availability of reliable language data and its adoption in everyday digital services.
