Orange Releases Five Open-Weight AI Models for Wolof Language
The initiative aims to address the underrepresentation of African languages in AI technologies and support the development of more inclusive digital services.
Orange has released five fine-tuned, open-weight large language models (LLMs) for Wolof, a language spoken by millions of people across West Africa, as part of efforts to expand the representation of African languages in artificial intelligence (AI).
The models are available free of charge for non-commercial use on Hugging Face, providing developers, researchers and public institutions with resources to build and test digital services that can understand and generate written Wolof.
The initiative aims to address the underrepresentation of African languages in AI technologies and support the development of more inclusive digital services. Potential applications include conversational assistants, frequently asked questions (FAQ) tools and guided digital services.
Orange said the models were developed and fine-tuned by its in-house AI teams, with five different model sizes to give developers flexibility in adapting them to specific applications.
The company also plans to join a global coalition of more than 60 organisations working to increase linguistic diversity in AI models. Through its participation, Orange will contribute its AI expertise and local knowledge across 18 markets in the Middle East and Africa.
“At Orange, we believe AI development must go hand in hand with a commitment to making it accessible to as many people as possible, while respecting local needs and ensuring it is developed and deployed responsibly. By drawing on our deep in-house expertise to fine-tune AI models, we can help harness their potential for the benefit of society.”
– Usman Javaid, Chief AI Officer, Orange
Orange Middle East Africa Deputy CEO and Director of Operations Ben Cheick Haidara said the models would help make digital services more accessible to people who communicate in local languages.
“The future of digital in Africa must undoubtedly be built around the languages people speak every day, easing communication and improving access to vital services. The publication of these models is more than a technical milestone; it delivers on a promise to make a real difference to people’s lives.”
– Ben Cheick Haidara, Deputy CEO and Director of Operations, Orange Middle East Africa
The Wolof models form part of Orange’s broader work to incorporate African languages into AI technologies, including Lingala, Swahili, Darija, Mooré and Dioula.
The accompanying training dataset is based on publicly available datasets automatically translated into Wolof and does not contain Orange corporate data. Both the models and training data are available on Hugging Face for non-commercial use, while commercial use may be subject to a separate agreement with Orange.
The release aligns with the company’s Trust the future strategy, which focuses on developing digital services that are more inclusive, accessible and trustworthy.

