I benchmarked GraphRAG on Groq vs Ollama. Groq is 90x faster.

The Comparison:

Ollama (Local CPU): $0 cost, 45 mins time. (Positioning: Free but slow)

OpenAI (GPT-4o): $5 cost, 5 mins time. (Positioning: Premium standard)

Groq (Llama-3-70b): $0.10 cost, 30 seconds time. (Positioning: The “Holy Grail”)

Live Demo:https://bibinprathap.github.io/VeritasGraph/demo/

https://github.com/bibinprathap/VeritasGraph