Low-latency local LLM runner via OpenJDK Panama FFM (Java 22)
I wanted to run AI from inside the JVM. I started out with the standard REST sidecar, ripped that out to use Project Panama (Foreign Function & Memory API) in the new JDK versions to interface directly with llama.cpp. I still wasn't happy with how that functioned, so I built libargus.cc to get a clean ABI to expose a structured API up in the JVM landscape. It still uses Project Panama to interface directly with llama.cpp, whisper.cpp, and ggml compute graphs. I have zero-allocation on the hot paths, memory segments for prompts and tokens are allocated once inside confined Arenas. Raw…
In plain words
This Java library enables running large language models directly within the JVM using OpenJDK 22's Panama FFM API, eliminating the need for separate REST services. It interfaces with llama.cpp and whisper.cpp through a custom C library (libargus.cc) that provides a structured API while maintaining low latency through zero-allocation hot paths and pre-allocated memory arenas. The solution bundles pre-compiled native binaries for straightforward deployment, allowing Java developers to integrate AI inference into their applications with minimal overhead.
written from the facts on this page · September 2026
From the sources
In the maker’s words, at launch
I wanted to run AI from inside the JVM. I started out with the standard REST sidecar, ripped that out to use Project Panama (Foreign Function & Memory API) in the new JDK versions to interface directly with llama.cpp. I still wasn't happy with how that functioned, so I built libargus.cc to get a clean ABI to expose a structured API up in the JVM landscape. It still uses Project Panama to interface directly with llama.cpp, whisper.cpp, and ggml compute graphs. I have zero-allocation on the hot paths, memory segments for prompts and tokens are allocated once inside confined Arenas. Raw pointers pass straight through down to the low C level. This avoids primitive array cloning and heap churn. I mapped out the native structures from llama.cpp and whisper.cpp while matching the compiler's padding to maintain safe memory access. I bundle pre-compiled native binaries in the jar for easy deployment. This execution engine provides the foundation I need for work I'm doing on a spatio-temporal memory layer (L-TABB) to replace RAGs. I'd love to get technical feedback to polish any issues while I continue working on the next layer. Deep-dives from anyone hacking on Project Panama or low-latency systems in modern JDK would be very appreciated! I'm much better with code than prose, so I'll let the code do most of the talking. Happy Hacking! /David Code: https://libargus.cc Project Landing Page: https://projectargus.cc
More ai this month
the category →
I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.
AI · 17d ago · simedw.com
Astute▲585Automate your B2B brand going viral, with new media creators
AI · 18d ago · company-app.joinastute.com
Hey HN, Henry from Cactus here! We previously released Cactus Needle, a 14MB agentic LLM for tool call, device use, and structured extraction for phones, wearables, smart homes, small robots and microcontrollers. We got really great feedback here, and have now incorporated the suggestions to release Needle 2. The whole model is a single 14MB binary that runs a full session in 28MB of RAM; 45m parameters at 2bit compression. Needle hits 500 tokens/sec decode speed on a Raspberry Pi 5, sits between 400-1,500 tokens/sec on VR devices like Meta Quest 3S and Apple Vision Pro, and ranges…
AI · 26d ago · cactuscompute.com
Launched alongside, July 2026
the whole month →- IR
I might be the only SRE on Earth with his own bowling center. It's a more in-depth gig than you'd think. My family and I bought an abandoned 8-lane bowling center in the rural mid-west. In our small town there weren't many recreation options for families. You've heard of a food desert? This is an R&R desert. It had been abandoned for a good reason. The roof leaks, the electrical system was constantly surging, and my 70-year-old bowling equipment (still) doesn't work perfectly. The system that keeps your score is particularly interesting to me. It's the thing you watch during your game, but…
Life & fun · Jul 2026
- EElevators▲1,680
Life & fun · Jul 2026 · john.fun
- 1W18 Words▲1,160
Life & fun · Jul 2026 · 18words.com
- BA
Over the past few months, our team has been building more and more slidedecks using web frontend technologies with coding harnesses like Claude Code, but a common complaint is to make even small edits we need to edit the code either manually or via the harness. To avoid this loop, I ended up creating Bento, a single HTML file with everything you need in a slide tool including animations and shared editing. There's no install or cloud login, everything works offline. The default deck is around 560 KB and it doesn't need to fetch anything once you got it. Open it in a browser and then you can…
Dev tools · Jul 2026 · bento.page
- GG
A few days ago I found myself trying out GLM 5.2 and was really positively impressed. The capabilities and security I was getting from this LLM are similar to those I've gotten from models like Claude or GPT, and this really surprised me. But then I thought, "I wonder how it would work on a normal computer like mine," and above all, "I wonder if it would work without going into OOM on a computer like mine." So I started working with the help of agents to test this possibility. I started converting the model to int4, understanding MTP usage, and if possible implementing DSA for long context.…
AI · Jul 2026 · github.com
