工程开发线上工程挑战奖金已截止

Africa Deep Tech Challenge 2026

Africa Deep Tech Challenge 2026 由 Africa Deep Tech Foundation 主办。报名截止 8月25日（已截止）。举办地区：全球。参赛对象：开发者、学生、研究者。奖励：现金奖金 16,500 美元以上，另有 GPU 额度和晋级机会

主办方：Africa Deep Tech Foundation· 来源智能体赛事网络发现 · 最近核验 8/24 15:22

前往官网报名报名已截止

赛事介绍

参赛团队需构建无需云依赖、可在标准 8GB 集成显卡笔记本运行的端到端语言模型，覆盖包括自主 AI 智能体在内的应用领域。

关键时间

报名6月17日 - 8月25日

赛事6月17日 - 8月25日

参赛条件与赛事信息

参赛对象: 开发者 · 学生 · 研究者
举办地区: 全球原始标注：全球线上（主题聚焦非洲场景）
报名截止: 8月25日（已截止）
奖励: 现金奖金 16,500 美元以上，另有 GPU 额度和晋级机会
赛事形式: 线上工程挑战
落地要求: 无（可远程）
报名入口: https://adtc-2026.devpost.com/

参赛资格

官方标注里没有排他的身份要求，符合赛事对象即可报名。

接受身份: student

报名要求

构建无云依赖的端到端本地语言模型
面向公布的标准笔记本硬件进行基准测试
提交开源 GitHub 仓库、项目报告和不超过 2 分钟视频
来源证据：https://devpost.com/hackathons

同类赛事

数据来源

本页信息来自公开来源（智能体赛事网络发现），最近核验于 8/24 15:22。赛事规则以主办方官方页面为准。

官方页面原文摘录

展开原文（6484 字）

About the challenge Africa is not excluded from the language model revolution. The bottleneck is no longer research or raw model availability - it is access economics. Cloud-hosted LLMs require API fees, stable fiber, and sustained electricity. For a university student in Lagos, an extension officer in Arusha, or a small-business owner in Dakar, these are not minor frictions - they are blockers. Africa Deep Tech Challenge 2026 (inspired by the Africa AI XPrize ) is an engineering-first competition to make useful language-model applications run well on the computers Africans already own: mid- and low-end commodity laptops. We are not building for exotic edge silicon. We are building for the 8 GB laptop with integrated graphics - the machine sitting on millions of desks, in classrooms, clinics, and corner shops across the continent. You could be the one to democratize access to low cost inference for the continent. This is an applied systems engineering contest. Hitting genuinely useful performance on commodity hardware requires optimization across the full stack: model selection and fine-tuning, quantization and compilation, memory and cache management, retrieval-augmented generation over local corpora, and application UX that makes such a model feel responsive. Get started Register on DevPost for the Africa Deep Tech Challenge 2026 Form or join a team of 1 to 3 people and pick your problem domain Clone the submission template Test your model locally by cloning the model profiler . Ensure your model and submission formats are perfectly compatible with the ADTC evaluation pipeline. The local profiler allows you to run latency, throughput, memory, and CPU checks directly on your target hardware. Study Frequently Asked Questions Apply for up to 5 hours of GPU credits courtesy of Udutech to train your model. Check out the rules and submission requirements, then start building! Requirements What to Build Participants build a working, end-to-end, on-device Language Model that runs without cloud dependencies on the ADTC Standard Laptop (defined below). The model must address one of the published problem domains Each team selects one primary domain. These are the ADTC standardized benchmarking domains. Math & Scientific Reasoning - problem solving, proof assistance, scientific question-answering, and quantitative reasoning tasks. Healthcare & Medical - clinical information, medical Q&A, triage support, and patient education. Agriculture - crop, livestock, weather, and market advisory for farmers and extension officers. Creative Writing - story generation, editing assistance, poetry, and narrative support across languages. Coding Assistants - code generation, debugging help, and programming tutoring across common languages. Corporate / Enterprise - knowledge-work productivity: summarization, drafting, and analysis for small and medium enterprises. Autonomous AI Agents - local orchestration, task automation, and privacy-focused workflow management using decentralized tools and messaging interfaces. All submissions are evaluated against a single, published reference hardware profile. This eliminates fragmentation and gives every participant a target they can design for. Participants may develop on any hardware, but final benchmarks and audits are reported against the Standard Laptop profile. Component ADTC Standard Laptop CPU Intel Core i5 10th–12th gen OR AMD Ryzen 5 3000–5000 (x86-64) RAM 8 GB DDR4 Graphics Integrated only (Intel UHD / Iris Xe or AMD Radeon integrated). No discrete GPU. Storage 256 GB SSD OS (reference) Ubuntu 22.04 LTS Representative Price $400–$500 new / $150–$250 refurbished What to Submit Open Source Github repo that leverages 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) Leaderboard Scoring S total = 0.50⋅ S acc +0.30⋅ S perf +0.20⋅ S eff − P thermal Component Weight Formula Notes S acc 50% Weighted average of model response scored between 0 and 100 by a Judge. Weighted combination of automated benchmark scores and qualitative assessment of model prompt responses by the judge panel. S perf 30% 100 × (TPSact ÷ TPSmax) TPS_REFERENCE = 15.0 provisional S eff 20% S eff = 100 × ((7 GB − Peak RAM) ÷ 7 GB) Rewards lower RAM usage. The less memory consumed relative to the 7 GB budget, the higher the score . Peak RAM = 7 GB P thermal -10 points -10 if throttled or temp > 85°C Else 0 Hackathon Sponsors Prizes $ 16,500 + in prizes + other prizes Grand Prize $ 8,000 in cash 1 winner Second Place $ 4,000 in cash 1 winner 3rd Place $ 3,000 in cash 1 winner Best African Use Case $ 1,500 in cash 1 winner Finalist Stipends 10 winners $250 (In GPU Credits) for up to 10 finalists Direct pathway into the Africa AI XPrize Semifinalist Stipends 20 winners $50 (In GPU Credits) for up to 20 semi finalists Devpost Achievements Submitting to this hackathon could earn you: Judges Oji Udezue Yannick Djoumbou Feunang Oluwatobi Oyinlola Christine Abernathy Omoju Miller Peter Ing Mbangula Lameck Amugongo Ola Fadiran Houda Ghozzi Judging Criteria Model Accuracy & Quality A combination of multiple-choice benchmarks and qualitative evaluations that includes accuracy of prompts, quality of documentation Model Throughput Performance Evaluated relative to the maximum observed tokens per second Model Efficiency Rewards lower RAM utilization profiles relative to the maximum 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 or sandbox execution crash results in disqualification Questions? Email the hackathon manager Invite others to compete Hackathon sponsors This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.