Orange is expanding its partnership with OpenAI to deploy two advanced open-weight reasoning models across its data centers, aiming to accelerate AI solutions development in its 26-country footprint, particularly in Africa.

The telecom operator said the new models — gpt-oss-120b and gpt-oss-20b — will allow greater control over deployment in a variety of environments, enabling it to host AI workloads locally while safeguarding sensitive data and meeting regulatory requirements across Europe, the Middle East, and Africa.

A key focus will be integrating African regional languages into large language models. By fine-tuning OpenAI’s open-weight models with local language data, Orange aims to let customers across its 18 African markets interact in their native languages with customer service, sales, and marketing platforms. The company plans to make these customized models available free to local governments for use in public services.

The deployment will also support applications such as natural-language voice assistants, chatbots, smarter network automation, advanced voice recognition, and AI-based analysis of sensitive data for enterprise clients.

“This strategy drives new use cases to address sensitive enterprise needs, help manage our networks, enable innovative customer care solutions including African regional languages, and much more,” said Steve Jarrett, Orange’s chief AI officer.