None of the Firebase data stores use graph database technology. Both Firestore and Realtime DB are NoSQL JSON document stores. However, it is possible to build a graph data model in Firestore. The article Graph Data with Firebase is a description of how to achieve this and there are many others.
Graphs, quite simply, are a way to model complex relationships between many objects. A graph is a collection of nodes (also known as vertices) and edges (also known as links). A node is merely an abstract data point and it can represent anything such as a person, a computer, a building, or an intersection. An edge connects two nodes and can optionally be directional. Information only flows one way.
The challenge is querying your graph. For some use cases, Firestore queries might be sufficient. However, powerful querying languages like Gremlin or Cypher are only available on true graph databases like Amazon Neptune or Neo4j.
Also, note that GraphQL just provides a query language for existing APIs and is completely unrelated to graph databases or data models.