Open Source AI Hackathon: Build Freely, Deploy Anywhere

The Pitch

The era of black-box AI controlled by a handful of providers is ending. Regulation is catching up, users are waking up, and open-source models have finally caught up to — and in many cases surpassed — their proprietary counterparts. This hackathon asks one question: What can you build when AI is open, inspectable, and yours to deploy?

Participants will have 7 hours to build impressive, functional applications powered entirely by open-source models — run them locally, on-prem, in the cloud, or on managed open-weight platforms. Your stack, your choice. The only rule: no closed-source frontier APIs (OpenAI, Anthropic, Google’s proprietary models, etc.) doing the heavy lifting.

Why This Theme, Why Now

The regulatory wave is here.

Norway’s dedicated AI Act is scheduled to enter into force in mid-2026 (CMS), implementing the EU AI Act’s risk-based framework. The GDPR, already in force through Norway’s Personal Data Act, requires purpose limitation and data minimisation for any AI system processing personal data (Chambers and Partners). For organisations navigating these requirements, open-source models offer the transparency and control that closed APIs can’t match — open weights mean you can audit, fine-tune, and deploy on your own terms.

The models are ready.

Open-source models have crossed a quality threshold. Fine-tuned small models (1B–30B parameters) now handle domain-specific tasks — legal Q&A, medical triage, code review, document processing — at quality levels that were unthinkable two years ago. Tools like Ollama, llama.cpp, vLLM, and managed open-weight providers make deployment trivial — locally or in the cloud.

The open-source shift is underway.

Open-weight models are powering everything from on-device assistants in phones and laptops to the fastest-growing startups on the internet. The next generation of tools — note-taking, health tracking, financial planning, agentic workflows — will be built on open foundations. Builders who understand this paradigm now will define it.