Lagos, Nigeria – June 2025 — Tech giant Meta has officially launched the Llama Impact Accelerator Program, a regional initiative aimed at supporting startups across Nigeria, Kenya, Senegal, and South Africa that are building impactful solutions using its open-source Llama large language models (LLMs).
The program, which is now open for applications, is designed to help African startups develop scalable, AI-powered solutions that address urgent challenges in sectors such as healthcare, agriculture, education, public safety, and public services.
“This initiative is about equipping Africa’s next generation of innovators with the tools, mentorship, and funding they need to build ethical and locally relevant AI applications,” Meta stated in the program announcement.
Key Dates
- Applications Open: May 26, 2025
- Deadline: June 27, 2025
- Accelerator Launch: August 2025
- Demo Days: September – October 2025
- Post-Program Support: October 2025 – March 2026
Benefits for Startups
- Up to $25,000 in equity-free funding for top teams
- Technical workshops and hands-on training with Llama AI tools
- Mentorship from global industry leaders and AI experts
- Post-program visibility, networking, and investor engagement
- Eligibility to pitch for $100,000 in additional funding at the AMET Llama Summit
Startups must be based in one of the four eligible countries, have a working prototype or MVP, and demonstrate strong technical capability and a clear social impact. Each country has its own thematic focus, with a “Wild Card” category open for exceptional innovations outside the listed sectors.
To apply, startups should visit the country-specific application portals:
Meta’s Llama models, known for their efficiency and accessibility, are designed to bring the power of AI to a broader range of developers and researchers. This accelerator marks a major step in Meta’s commitment to supporting inclusive AI development across the Global South.
For more details or to attend the June 23 info session, visit the official Llama Impact Accelerator page.
