A tiny LLM running at 21,000 tok/s on a $250 FPGA (Live Demo)
A 3.16M-parameter INT4 transformer running entirely in the on-chip memory of a Xilinx Kria KV260. Zero DRAM in the token loop, 59,965 tok/s on the fabric, bit-exact. Chat with it live.
In plain words
A 3.16-million-parameter language model runs entirely on the on-chip memory of a $250 Xilinx Kria KV260 FPGA, achieving 59,965 tokens per second without using external DRAM. The INT4 transformer operates as a live chatbot, with the model's weights and inference computations contained fully within the FPGA's reconfigurable logic. This project demonstrates that practical language model inference is possible on consumer-grade hardware through efficient quantization and specialized hardware deployment.
written from the facts on this page · September 2026
From the sources
A 3.16M-parameter INT4 transformer running entirely in the on-chip memory of a Xilinx Kria KV260. Zero DRAM in the token loop, 59,965 tok/s on the fabric, bit-exact. Chat with it live. I was so impressed by the chatjimmy.ai demo by Taalas, I wanted to see what I could squeeze inside the fabric fabric: The FPGA's reconfigurable logic. Same thing as 'PL'. Where the whole model runs here. of a $250 FPGA FPGA: Field-Programmable Gate Array: a chip full of reconfigurable logic you wire into a custom digital circuit, instead of running software on a fixed CPU. . By not using the (4 GB available) DDR DDR: The off-chip DRAM (the board's main memory), ~20 GB/s, shared by both the CPU and the fabric.…from mikeayles.com
Does the same job
all alternatives →- IBI built a tiny LLM to demystify how language models workApr 2026 · github.com · ▲915
Built a ~9M param LLM from scratch to understand how they actually work. Vanilla transformer, 60K synthetic conversations, ~130 lines of PyTorch. Trains in 5 min on a free Colab T4. The fish thinks the meaning of life is food. Fork it and swap the personality for your own character.
- OSOpen-source engine running Gemma 4 26B in 2 GB RAM on any M-series MacJul 2026 · github.com · ▲919
Hi HN, I built a specialized inference engine for running 4-bit Gemma 4 26B-A4B-IT on any M-series Mac using about 2 GB of RAM. It is called TurboFieldfare and is written in Swift and Metal. I have always adored on-device AI. It feels like magic that you can run a powerful NN on your Mac or iPhone. So I wanted to push the limits a bit and run a model whose weights don’t fit in memory. The model’s 4-bit quantized weights occupy roughly 14 GB, which makes running it with conventional inference tools almost impossible on an 8 GB or even 16 GB Mac once the OS, applications, and KV cache are…
Needle2: 14MB agentic LLM for phones, wearables, smart home and robots27d ago · cactuscompute.com · ▲537Hey 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…
- TNThree new Kitten TTS models – smallest less than 25MBMar 2026 · github.com · ▲561
Kitten TTS (https://github.com/KittenML/KittenTTS) is an open-source series of tiny and expressive text-to-speech models for on-device applications. We had a thread last year here: https://news.ycombinator.com/item?id=44807868. Today we're releasing three new models with 80M, 40M and 14M parameters. The largest model (80M) has the highest quality. The 14M variant reaches new SOTA in expressivity among similar sized models, despite being <25MB in size. This release is a major upgrade from the previous one and supports English text-to-speech applications in…
- TVTiny-vLLM – high performance LLM inference engine in C++ and CUDAMay 2026 · github.com · ▲205
- 1B1-Bit Bonsai, the First Commercially Viable 1-Bit LLMsMar 2026 · prismml.com · ▲430
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 · 27d ago · cactuscompute.com
Launched alongside, August 2026
the whole month →- TL
Life & fun · 10d ago · louisabraham.github.io
- SA
Hello HN! I found that picking out plausible but diverse skin tones for my digital art and game development projects was kind of difficult, and I got curious about if there was a way to define a color space that made it easy. I've built a color picker and procedural generation algorithm based on the space as well as a bunch of other fun js features and demos throughout the page that use the equations. If you find it interesting, I have lots of explanations of how I built it and what properties the space has. The methodology might be a bit shaky, but hopefully the result is as helpful for…
Life & fun · Aug 2026 · toneyalexander.github.io
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