Knowledge bases (KBs) are a cornerstone of the Semantic Web, yet they still struggle with scale and scope, and their construction and curation still involve a lot of manual effort. Large language models (LLMs) have recently emerged as powerful tools for a range of tasks, yet their potential for automated KB construction is still poorly understood. In this demonstrator, we showcase GPTKB, a methodology and KB entirely built from GPT-4.1. GPTKB is constructed by massive-recursive LLM knowledge materialization [1], using over 9M API calls for $14,000 to construct a 100M-triple knowledge base with over 6M entities. Our demonstration focuses on two use cases: (i) Link-based KG exploration and (ii) SPARQL-based analysis and comparison to Wikidata. The GPTKB demonstrator is accessible at https://gptkb.org.