Africa Deep Tech Challenge 2026

Overview

The Africa Deep Tech Challenge 2026 addresses the critical issue of accessibility to advanced language models in Africa, where cloud-hosted LLMs are often constrained by API fees, stable internet, and consistent electricity. This challenge focuses on developing practical, on-device language model applications that can run efficiently on the common hardware found across the continent: mid- and low-end commodity laptops with 8 GB RAM and integrated graphics. Inspired by the Africa AI XPrize, this competition is a systems engineering challenge. Participants are tasked with optimizing the entire stack—including model selection, fine-tuning, quantization, compilation, memory management, and retrieval-augmented generation—to achieve useful performance on hardware representative of what millions of Africans already own. The goal is to democratize access to low-cost AI inference, enabling solutions in vital sectors like healthcare, agriculture, and education without reliance on expensive cloud infrastructure. Teams will select a problem domain from categories such as Math & Scientific Reasoning, Healthcare & Medical, Agriculture, Creative Writing, Coding Assistants, Corporate/Enterprise, or Autonomous AI Agents. Submissions will be evaluated against a standardized laptop profile to ensure fairness and focus on practical, resource-constrained performance. This challenge is for early-stage innovators with projects less than 12 months old and minimal funding, fostering local innovation to solve local problems.

What to build

Participants must build a working, end-to-end, on-device Language Model that operates without cloud dependencies on the ADTC Standard Laptop hardware profile (Intel Core i5 10th–12th gen or AMD Ryzen 5 3000–5000, 8 GB DDR4 RAM, integrated graphics, 256 GB SSD, Ubuntu 22.04 LTS). The model must address one of the seven published problem domains: * Math & Scientific Reasoning: for problem-solving, proof assistance, scientific Q&A, and quantitative reasoning. * Healthcare & Medical: for clinical information, medical Q&A, triage support, and patient education. * Agriculture: for crop, livestock, weather, and market advisory. * Creative Writing: for story generation, editing assistance, poetry, and narrative support across languages. * Coding Assistants: for code generation, debugging, and programming tutoring. * Corporate / Enterprise: for knowledge-work productivity like summarization, drafting, and analysis for SMEs. * Autonomous AI Agents: for local orchestration, task automation, and privacy-focused workflow management.

Prizes

Key dates

• Contest OpensTuesday, June 16, 2026
• Gate 1 Deadline (Proposals + Prototypes)Tuesday, August 25, 2026
• Semifinalists AnnouncedTuesday, September 8, 2026
• Semifinalist Submission DeadlineTuesday, September 22, 2026
• Finalists AnnouncedTuesday, September 29, 2026
• Live Defense & AwardsSaturday, October 17, 2026

Eligibility

This challenge is strictly for early-stage builders. Entrants must formally declare that their project or company is less than 12 months old and that their product is in the ideation, conceptual, or early prototype phase (not commercially launched). Any team found to have misrepresented their stage, age, or funding will be immediately disqualified. Participants must be above the legal age of majority in their country of residence and reside in one of the listed African countries. Teams can be individuals or groups of up to 3 members. No prior startup or funding status is required. Incorporations, formal registrations, or organized teams must not exceed 12 months of existence as of June 16, 2026. Submissions must be at the ideation, conceptual, or early proof-of-concept (PoC) stage; fully commercialized products, active monthly recurring revenue (MRR), or mature MVPs deployed for more than 6 months will result in disqualification. Participating teams must not have raised more than $25,000 USD in external dilutive capital or non-dilutive grants. All submissions must be original work, and open-source tools or libraries must be clearly cited.

How to enter

1. Register on DevPost for the Africa Deep Tech Challenge 2026.
2. Form or join a team of 1 to 3 people and pick your problem domain.
3. Clone the submission template from GitHub.
4. Test your model locally using the model profiler to ensure compatibility with the ADTC evaluation pipeline.
5. Apply for up to 5 hours of GPU credits from Udutech for model training.
6. Review the rules and submission requirements, then start building.
7. Submit your open-source GitHub repo and comprehensive project report by the Gate 1 Deadline (August 25, 2026).

Submission requirements

• Open Source GitHub repository leveraging the approved ADTC 2026 Report Template.
• A comprehensive project report including: - Problem definition and context - Identified constraints (e.g., power, data, compute, connectivity) - Documentation of design alternatives and final decisions - Tools used and why they were chosen - Performance tests and benchmarks - Screenshots or short videos showing your build in action
• A short video (max 2 minutes) explaining your solution and development journey.
• Updated repo, documentation, and video (if part of semi-final or final round).

Judging criteria

• Model Accuracy & Quality (50%): Weighted combination of automated benchmark scores and qualitative assessment of model prompt responses by the judge panel, including accuracy of prompts and quality of documentation.
• Model Throughput Performance (30%): Evaluated relative to the maximum observed tokens per second (TPS).
• Model Efficiency (20%): Rewards lower RAM usage relative to the 7 GB memory budget.
• African Use Case Bonus (Up to 10 extra points): Awarded for how applicable the model is to a real African use case.
• Hardware & Thermal Penalties (-10 points): Deducted if core/package temperature exceeds 85°C or if thermal throttling is flagged. OOM (Out Of Memory) or sandbox execution crash results in disqualification.

Sponsors & partners

Resources

• ADTC 2026 Submission Template
• ADTC Model Profiler
• Apply for GPU Credits (Udutech)
• Challenge Participation Agreement
• Common AI Model Formats
• A Practical Guide to GGUF Quantization Selection
• Official Ollama Documentation & Setup Guide
• llama.cpp Official Build Guide (Linux)
• llama.cpp Discussions: Optimizing CPU-only and 8GB RAM Performance
• Monitor & Control CPU Temperature on Ubuntu 22.04
• Build Small Hackathons with Cohere Models
• Masakhane: African NLP Benchmarks and Datasets
• LLM Benchmark for Throughput via Ollama
• llama-bench: Syntax, Usage and Documentation
• Africa Deep Tech Challenge 2026 Discord Channel
• Africa Deep Tech Challenge 2026 Intro: The Laptop LLM Challenge (YouTube)
• Watch Knowledge Session Recording: The Last Mile of AI: Designing for Real-World Constraints (YouTube)
• Zoom Registration for Knowledge Session