Skip to main content

Quarkus

The Quarkus LangChain4j extension seamlessly integrates with the Quarkus programming model and existing Quarkus runtime components.

The extension offers the following advantages over using the vanilla LangChain4j library in Quarkus:

  • Integration with the Quarkus programming model
    • A new @RegisterAiService annotation for declarative AI services
    • Injectable CDI beans for the LangChain4j models
  • Ability to compile to a GraalVM native binary -Standard configuration properties for configuring models
  • Built-in observability (metrics, tracing, and auditing)
  • Build-time wiring. Doing more at build-time reduces the footprint of the LangChain4j library and enables build-time usability hints.

Dev UI​

In Dev mode, the quarkus-langchain4j project provides several pages in the Dev UI to facilitate LangChain4j development:

  • AI Services page: provides a table of all AI Services detected in the application along with a list of tools that they are declared to use.
  • Embeddings store access: Allows embeddings to be added to the embeddings store and searched.
  • Tools page: provides a list of tools detected in the application.
  • Chat page: allows you to manually hold a conversation with a chat model. This page is only available if the application contains a chat model.
  • Images page: allows you to test the outputs of image models and tune its parameters (for models which support it).
  • Moderation page: allows you to test the outputs of moderation models - you submit a prompt and receive a list of scores for each appropriateness category (for models which support it).

For more detailed explanation of the extension features, see the Quarkus documentation for the langchain4j extension.