Welcome, builders. This page has everything you need to navigate the GraphRAG Inference Hackathon. Bookmark it, share it with your team, and use it as your single source of truth.
LLMs are spreading rapidly across every industry, and as they do, token consumption is exploding. Companies pay more, wait longer, and hit context limits faster every quarter. This isn't an academic exercise. It's a real, growing pain in production AI today.
The GraphRAG Inference Hackathon by TigerGraph is a beginner-friendly, online hackathon where you'll prove that graphs make LLM inference faster, cheaper, and smarter than vector-based RAG alone. You'll build three pipelines side-by-side (raw LLM, Basic RAG, GraphRAG) and let the numbers tell the story.
The headline metric: token reduction with maintained accuracy. Show us how much GraphRAG cuts down without dropping answer quality, and you'll have built the kind of project that gets hiring managers, AI teams, and the open-source community paying attention.
| Benefit | What it means |
|---|---|
| Beginner-friendly | No prior TigerGraph or graph experience required. Python and basic LLM API usage is enough. |
| Open globally | Students, professionals, solo builders, and teams of up to 5 welcome. |
| $700 prize pool | Cash prizes plus mentoring from the TigerGraph engineering team for the Top 15. |
| Make an impact | Your findings and contributions will help solve a real and growing industry-wide problem: token efficiency in production AI. |
| Free to enter | No fees, no catches. |
| Free TigerGraph stack access | Full free access to Savanna ($60 in credits, more if needed), Community Edition, MCP, and the GraphRAG repo. |
| Direct engineer access | 1:1 mentoring sessions, office hours, and WhatsApp support throughout. |
You need two things before diving in: a TigerGraph environment and the GraphRAG repo.
Option A: TigerGraph Savanna (recommended). Web-based, zero installation. Sign up at tgcloud.io and get ~$60 in free credits instantly (more if needed).
Option B: Community Edition. Free, runs locally. Download from dl.tigergraph.com.
git clone <https://github.com/tigergraph/graphrag.git>
This repo is the foundation for Pipeline 3. You'll either use it as-is or build on top of it (more on that in the Problem Statement below).
You'll need access to any LLM provider of your choice for all three pipelines and for the accuracy evaluation. Use whichever you're comfortable with. Most major providers offer free tiers that are more than enough for hackathon-scale usage.