Press kit

Material for newsletter editors, journalists, conference organisers and bloggers who write about LLM4S. You may quote the descriptions below as they are, or shorten them. Every fact on this page can be checked in the repository; it was last checked on 2026-11-08, for release 0.5.1.

  2. Descriptions
    1. 50 words
    2. 150 words
  3. Fact sheet
  4. Try it
  5. Contact
Asset File Use
Logo, mark and name, on a light background llm4s_logo.png (576×576 PNG) Articles, slides, listings
Share card default.png (1200×630 PNG) Link previews, banners

The logo is the red mark (three slanted bands beside a brain outline with connected nodes) above the black name LLM4S. Write the name as LLM4S in running text, and llm4s only in code, package and artifact names. Do not recolour, stretch or redraw the logo.

Vector (SVG) versions of the logo for light and dark backgrounds, and a separate wordmark, are not published yet; until they are, use the PNG above and ask the contact below for a larger size.

Descriptions

50 words

LLM4S (Large Language Models for Scala) is an open-source framework for building LLM applications on the JVM. One typed client reaches OpenAI, Anthropic, Gemini, Ollama, AWS Bedrock and other providers; its agent runtime has tools, guardrails and handoffs. Errors are values, not exceptions. MIT-licensed Scala 3, with Java and Kotlin APIs.

150 words

LLM4S (Large Language Models for Scala) is an open-source framework for building applications on large language models, on the JVM. It gives Scala, Java and Kotlin developers one typed client for many providers: OpenAI, Azure OpenAI, Anthropic, Google Gemini and Vertex AI, AWS Bedrock, Ollama for local models, and any OpenAI-compatible endpoint. Each provider is a separate module, found through a documented provider interface.

On top of the client, LLM4S provides an agent runtime with tools, guardrails, handoffs, streaming and durable threads; retrieval-augmented generation over pgvector, Qdrant, SQLite, Elasticsearch, OpenSearch and Redis; a Model Context Protocol client and server; and tracing through Langfuse and OpenTelemetry. Failures come back as typed values instead of exceptions.

LLM4S is MIT-licensed and developed in public on GitHub since January 2025. It is community-led, maintained by a small core team and stewarded by the LLM4S Foundation, which is funded through Open Collective.

Fact sheet

   
Name LLM4S (Large Language Models for Scala)
What it is An open-source library for building LLM applications on the JVM: provider clients, agents, tools, RAG, MCP, tracing
Website llm4s.org
Source github.com/llm4s/llm4s
Licence MIT (LICENSE)
First release 0.1.0, April 2025
Latest release 0.5.1 (release notes, news)
Maven coordinates org.llm4s group on Maven Central, for example org.llm4s:llm4s-core_3
Language and runtime Scala 3.7.1, JDK 21 or later (bytecode for JDK 21)
Java llm4s-java-api (a Java facade over the client and agent), a Spring Boot starter and Reactor support
Kotlin llm4s-kotlin-api: coroutines and a DSL for tools and agents, built with Kotlin 2.0.21 for JVM 21; published from 0.5.2
Chat providers OpenAI, Azure OpenAI, Anthropic, Google Gemini, Vertex AI, AWS Bedrock, Ollama, DeepSeek, Mistral, Cohere, xAI, Perplexity, OpenRouter, Requesty, Z.ai, IBM watsonx.ai (beta), and any OpenAI-compatible endpoint by configuration
Embedding providers OpenAI, Voyage AI, Jina AI, Cohere, Ollama
Effect and stream libraries cats-effect and fs2, ZIO 2, Apache Pekko Streams
Governance Maintainers decide by lazy consensus, majority vote for contested changes; public API changes go through RFCs (GOVERNANCE.md, MAINTAINERS.md)
Maintainers Kannupriya Kalra and Rory Graves (people)
Stewardship and funding The LLM4S Foundation, funded through Open Collective (Foundation)
Community Discord, GitHub issues, Google Summer of Code (GSoC)

Try it

Readers can try LLM4S without an API key, against a small local model:

Contact

For interviews, quotes, logo files or corrections, email kannupriya@llm4s.org or ask on Discord.