# tjake/Jlama

Jlama is a modern LLM inference engine for Java

Repository: https://github.com/tjake/Jlama
Canonical: https://ross.abutalabs.com/products/jlama
Language: Java
License: Apache-2.0
License Family: permissive
Topics: ai, java, llm, simd, transformers, gpt, llama, openai, huggingface, genai
Last push: 2025-10-12T02:45:08+00:00

## Health v2 (maintenance only)
Score: 51/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 46, release rhythm 40, longevity 80
- inputs: {"age_days": 1129, "days_push": 325, "days_rel": 609, "gap_med": 6.5, "n_releases_24m": 7}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1303, forks 164 (observed 2026-08-28T04:04:18.327031+00:00)

## What it is
Jlama is a modern LLM inference engine written in Java, supporting popular transformer models like Llama, Mistral, Gemma, and Qwen with quantization and distributed inference. It ships as an embeddable library plus a CLI that can serve an OpenAI-compatible REST API with a chat UI.

## Use cases
- run llm inference in a java application
- serve an openai-compatible api for local models
- chat with huggingface models from the command line
- run quantized llama or mistral models locally
- generate embeddings with a java library
- distributed inference across multiple machines

## When to choose
- your application is on the JVM and you want built-in LLM inference without Python
- you want an OpenAI-compatible local REST API with a chat UI
- you need quantized model support (Q8/Q4) in pure Java
- you want distributed inference across a cluster of JVMs

## When to avoid
- you need maximum inference throughput available in GPU-native runtimes like vLLM or llama.cpp
- you need broad model architecture support beyond the supported families
- you are not on Java 20+ since the engine requires modern JDK features

## Facets
- artifact type: library
- maturity: active
- function: llm-inference, cli, http-server, machine-learning, transformers
- domain: large-language-models, artificial-intelligence, machine-learning, developer-tools
- platform: jvm, cli, cross-platform
- tags: llm-inference, java, simd, quantization, openai-compatible-api, huggingface, distributed-inference, panama-vector, safetensors

## Member repositories
- tjake/Jlama (main) score 51

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:18.327031+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:51:14.206738+00:00, confidence not recorded.
  - readme: https://github.com/tjake/Jlama (fetched 2026-08-28T04:04:18.327031+00:00, sha 11519ad50bb4)
- Data as of 2026-08-30T08:39:29.467469+00:00.