Why African Universities Should Prioritize Open-Source Large Language Models: Evidence from Computing Career Guidance Across Ten  Countries

African universities face a strategic decision: invest in expensive proprietary large language models (LLMs) or adopt emerging open-source alternatives for educational applications. This paper presents, to our knowledge, the first cost-benefit analysis and implementation framework for institutional decision-making in computing career guidance. Based on a comprehensive evaluation of six LLMs across ten African countries, leading open-source models achieved stronger performance (4.25–4.47 out of 5) compared to proprietary alternatives (3.46–3.90 out of 5), while reducing costs by up to 70×. Beyond cost savings, open-source models allow African universities to develop locally adapted AI systems by fine-tuning with local educational data and languages. The paper provides actionable recommendations for universities, policymakers, and businesses, demonstrating how the strategic adoption of open-source technologies can enhance academic outcomes while affording Africa digital sovereignty and localized AI capacity, which aligns with the African Union's Continental AI Strategy and the United Nations' Sustainable Development Goals 4 and 9.