Alternatives
Products that do what The fastest way to run Mixtral 8x7B on Apple Silicon Macs does
I’d originally launched my app: Private LLM[1][2] on HN around 10 months ago, with a single RedPajama Chat 3B model. The app has come a long way since then. About a month ago, I added support for 4-bit OmniQuant quantized Mixtral 8x7B Instruct model, and it seems to outperform Q4 models at inference speed and Q8 models at text generation quality, while consuming only about 24GB of RAM[3] at 8k context length. The trick is: a) to use a better quantization algorithm and b) to use unquantized embeddings and the MoE gates (the overhead is quite small). Other notable features include many more…
- 1SU
Here's a project I've been working on for the last few months. It's a new (I think) algorithm, that allows to adjust smoothly - and in real time - how many calculations you'd like to do during inference of an LLM model. It seems that it's possible to do just 20-25% of weight multiplications instead of all of them, and still get good inference results. I implemented it to run on M1/M2/M3 GPU. The mmul approximation itself can be pushed to run 2x fast before the quality of output collapses. The inference speed is just a bit faster than Llama.cpp's, because the rest of implementation…
2024 · asciinema.org
- 2
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…
Jul 2026 · github.com
- 3IB
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.
Apr 2026 · github.com
- 4IM
Hi Hackers, Excited to share a macOS app I've been working on: https://recurse.chat/ for chatting with local AI. While it's amazing that you can run AI models locally quite easily these days (through llama.cpp / llamafile / ollama / llm CLI etc.), I missed feature complete chat interfaces. Tools like LMStudio are super powerful, but there's a learning curve to it. I'd like to hit a middleground of simplicity and customizability for advanced users. Here's what separates RecurseChat out from similar apps: - UX designed for you to use local AI as a daily driver.…
2024 · recurse.chat
- 5KR
I discovered that in LLM inference, keys and values in the KV cache have very different quantization sensitivities. Keys need higher precision than values to maintain quality. I patched llama.cpp to enable different bit-widths for keys vs. values on Apple Silicon. The results are surprising: - K8V4 (8-bit keys, 4-bit values): 59% memory reduction with only 0.86% perplexity loss - K4V8 (4-bit keys, 8-bit values): 59% memory reduction but 6.06% perplexity loss - The configurations use the same number of bits, but K8V4 is 7× better for quality This means you can run LLMs with 2-3× longer…
2025 · github.com
- 6GG
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.…
Jul 2026 · github.com
- 7WM
Try it out! https://glhf.chat/ Hey HN! We’ve been working for the past few months on a website to let you easily run (almost) any open-source LLM on autoscaling GPU clusters. It’s free for now while we figure out how to price it, but we expect to be cheaper than most GPU offerings since we can run the models multi-tenant. Unlike Together AI, Fireworks, etc, we’ll run any model that the open-source vLLM project supports: we don’t have a hardcoded list. If you want a specific model or finetune, you don’t have to ask us for it: you can just paste the Hugging Face link in and…
2024 · glhf.chat
- 88F
Hi HN! I'm just sharing a project I've been working on during the LLM Efficiency Challenge - you can now finetune Llama with QLoRA 5x faster than Huggingface's original implementation on your own local GPU. Some highlights: 1. Manual autograd engine - hand derived backprop steps. 2. QLoRA / LoRA 80% faster, 50% less memory. 3. All kernels written in OpenAI's Triton language. 4. 0% loss in accuracy - no approximation methods - all exact. 5. No change of hardware necessary. Supports NVIDIA GPUs since 2018+. CUDA 7.5+. 6. Flash Attention support via Xformers. 7. Supports 4bit and 16bit…
2023 · github.com
- 9IB
I’ve been playing around with local LLMs for the past couple of months and decided to build something that can run on an iPhone. It’s a universal app built with SwiftUI and the excellent ggml library. The model is an SFT fine tuned and 4 bit quantised version of the RedPajama-INCITE-Chat-3B-v1 OSS LLM. It works reasonably well on recent-ish (~3 year old) iPhones, iPads and Macs. It was launched on the App Store yesterday[1] and Product Hunt today[2]. It seems to be reasonably ok at natural language interactions, but given its size, does pretty badly at coding and reasoning. Also, it…
2023
- 10
I built slotstream, a way to run Qwen3.8-Flash-Next 4-bit on a low-memory mac starting from 16GB, a 125B parameter model that would need 100GB+ memory/RAM, thanks to expert-offloading/ssd-streaming. Easy to install/update, and mac-native using MLX and Swift. It ships with auto-mode, which makes a good tradeoff between memory usage and speed. I'll be implementing and porting the MTP module for speculative decoding next Local models really are the future of computing!
5d ago · github.com
- 11WM
We wrote our inference engine on Rust, it is faster than llama cpp in all of the use cases. Your feedback is very welcomed. Written from scratch with idea that you can add support of any kernel and platform.
2025 · github.com
- 12SO
We built SwiftAI, an open-source Swift library that lets you use Apple’s on-device LLMs when available (Apple opened access in June), and fall back to a cloud model when they aren’t available — all without duplicating code. SwiftAI gives you: - A single, model-agnostic API - An agent/tool loop - Strongly-typed structured outputs - Optional chat state Backstory: We started experimenting with Apple’s local models because they’re free (no API calls), private, and work offline. The problem: not all devices support them (older iPhones, Apple Intelligence disabled, low battery, etc.). That…
2025 · github.com
- 13
Swiftlet is a Swift and Metal runtime that runs large Qwen Mixture-of-Experts models locally on Apple devices by streaming expert weights from storage, enabling 35B and 80B models to run with low RAM, including on iPhone. - leonickson1/Swiftlet
Aug 2026 · github.com
- 14
- 15
Fine-tune LLMs from one YAML. Layer streaming trains an 8B model on a 4 GB laptop GPU. - MakazhanAlpamys/Soup
Aug 2026 · github.com
- 16TV
May 2026 · github.com
- 17OR
Hi HN A few folks and I have been working on this project for a couple weeks now. After previously working on the Docker project for a number of years (both on the container runtime and image registry side), the recent rise in open source language models made us think something similar needed to exist for large language models too. While not exactly the same as running linux containers, running LLMs shares quite a few of the same challenges. There are "base layers" (e.g. models like Llama 2), specific configuration to run correctly (parameters, temperature, context window sizes etc). There's…
2023 · github.com
- 18
- 19
- 20R5
Hi HN, I built OpenGraviton, an open-source AI inference engine that pushes the limits of running extremely large LLMs on consumer hardware. By combining 1.58-bit ternary quantization, dynamic sparsity with Top-K pruning and MoE routing, and mmap-based layer streaming, OpenGraviton can run models far larger than your system RAM—even on a Mac Mini. Early benchmarks: TinyLlama-1.1B drops from ~2GB (FP16) to ~0.24GB with ternary quantization. At 140B scale, models that normally require ~280GB fit within ~35GB packed. Optimized for Apple Silicon with Metal + C++ tensor unpacking, plus…
Mar 2026 · github.com
- 21TL
2023 · tinyllms.vercel.app
- 22
- 23MG
Hello HN, I've been working on this project for a while, and it has been in an "open" beta for some time. I finally believe it's ready for its first release. I hope you like it. Here are some potential questions that may arise: 1. How does it compare to LM Studio? It's likely that if you're already using LM Studio, you'll continue to do so. This project is designed to be more user-friendly. 2. Is it open-source? No, it is not. 3. Does it use any open-source libraries? Yes, it uses llama.cpp and a few others, as indicated in the license information included with the application. 4. Why is not…
2023 · avapls.com
- 24BR
Check out this impressive project that enables running LLMs entirely in the browser using WebGPU. Key features: - Zero token costs, no cloud infrastructure required - Complete data privacy through local processing - Simple 3-line code integration - Built on MLC and Transformer.js The benchmarks show smaller models can effectively handle many common tasks. Currently the project roadmap includes: - No-code AI pipeline builder - Browser-based RAG for document chat - Analytics/logging - Model fine-tuning interface
2025 · github.com
Ranked by how close each launch is in meaning, then by votes. Refine with a description →