AI · alternatives · 2026

24 alternatives to Open-source engine running Gemma 4 26B in 2 GB RAM on any M-series Mac

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…

TurboFieldfare is an inference engine built in Swift and Metal that enables M-series Macs to run the Gemma 4 26B model using approximately 2 GB of RAM. It uses a specialized technique that keeps essential model… Below are 24 products that do a similar job, ranked by how close each is in meaning and then by launch-day votes.

1
We made our own inference engine for Apple Silicon▲186
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 · its alternatives →
2
Run an 80B Qwen in 4.3 GB of RAM on a Mac, and a 35B on an iPhone▲312
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 · its alternatives →
3
Running 104GB Qwen3.8-Flash-Next on 48GB Mac with at ~12 tok/s▲240
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!
Sep 2026 · github.com · its alternatives →
4
Google Gemma 4▲437
Google's most intelligent open models to date
Apr 2026 · blog.google · its alternatives →
5
Gemma 4 Multimodal Fine-Tuner for Apple Silicon▲235
About six months ago, I started working on a project to fine-tune Whisper locally on my M2 Ultra Mac Studio with a limited compute budget. I got into it. The problem I had at the time was I had 15,000 hours of audio data in Google Cloud Storage, and there was no way I could fit all the audio onto my local machine, so I built a system to stream data from my GCS to my machine during training. Gemma 3n came out, so I added that. Kinda went nuts, tbh. Then I put it on the shelf. When Gemma 4 came out a few days ago, I dusted it off, cleaned it up, broke out the Gemma part from the Whisper…
Apr 2026 · github.com · its alternatives →
6
Ollama v0.19▲410
Massive local model speedup on Apple Silicon with MLX
Apr 2026 · ollama.com · its alternatives →
7
Run TRELLIS.2 Image-to-3D generation natively on Apple Silicon▲202
I ported Microsoft's TRELLIS.2 (4B parameter image-to-3D model) to run on Apple Silicon via PyTorch MPS. The original requires CUDA with flash_attn, nvdiffrast, and custom sparse convolution kernels: none of which work on Mac. I replaced the CUDA-specific ops with pure-PyTorch alternatives: a gather-scatter sparse 3D convolution, SDPA attention for sparse transformers, and a Python-based mesh extraction replacing CUDA hashmap operations. Total changes are a few hundred lines across 9 files. Generates ~400K vertex meshes from single photos in about 3.5 minutes on M4 Pro (24GB). Not as fast as…
Apr 2026 · github.com · its alternatives →
8
Needle2: 14MB agentic LLM for phones, wearables, smart home and robots▲537
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…
Aug 2026 · cactuscompute.com · its alternatives →
9
Osaurus▲534
Open source agents that run 100% locally on your Mac
Jul 2026 · osaurus.ai · its alternatives →
10
Gemma 2▲279
Lightweight, state-of-the-art open models from Google
2024 · blog.google · its alternatives →
11
Cactus Hybrid: We taught Gemma 4 to know when it's wrong▲191
Hey HN, Henry & Roman here from Cactus. A small, on-device model is fast and private, but sometimes wrong, but frontier models are getting expensive pretty fast. So, we post-trained Gemma 4 E2B post-trained to know when it's wrong. Every response comes with a confidence score between 0 and 1. Developers can accept the on-device when it's high, hand off to a bigger cloud model when it's low. By routing only 15-35% of queries to Gemini 3.1 Flash-Lite, Gemma-4-E2B matches Gemini 3.1 Flash-Lite on most benchmarks. - ChartQA: 15-20% - LibriSpeech: 25-30% - MMBench, GigaSpeech, MMAU: 30-35% -…
Jul 2026 · github.com · its alternatives →
12
Running Gemma-4 26B at 124 tokens/SEC on a CPU, no GPU▲10
I wanted to know how fast a 26B mixture-of-experts model could run on a desktop CPU with no GPU. Got ~40 tok/s single-stream (lossless) and ~124 batched. The surprising part was the byte budget: for this model you compress the output head (32% of per-token bytes), not the experts (16%). The writeup has the bandwidth roofline and the dead-ends; the repo has the reproducible recipe. Happy to answer questions. Repo: https://github.com/arun-prasath2005/gemma4-cpu-moe
Jun 2026 · apeg.dev · its alternatives →
13
Inference Engine by GMI Cloud▲180
Fast multimodal-native inference at scale
Dec 2025 · gmicloud.ai · its alternatives →
14
I made a Gemma 4 Mac app that names screenshots with local AI▲7
I made my first macOS utility app that ships with a bundled Gemma 4 model, specifically the Gemma E4B one. It made my app DMG have 5.3 GB in size, but I think it is a small size for the power that this free local model can provide. It runs fine on CPU, but can also run on Apple Silicon GPU, although I did not notice any performance improvements with GPU (tested on a M5 chip). I think these local lightweight and multimodal models will open multiple possibilities for new software tools where privacy is essential.
May 2026 · snapname.app · its alternatives →
15
Cai▲179
Hi HN, Cai runs actions on your selection via ⌥C, pick an action (AI, shell, GitHub issue, Linear ticket), done. Bundled local model (MLX Community), create custom actions with prompts, scripts and more. Free, MIT. Built after my Raycast extension sat in review queue for 3+ months. Wanted no typing, no app switching, no cloud. Repo: https://github.com/cai-layer/cai Demo: https://youtu.be/LH2MYNQ_wsk
Apr 2026 · getcai.app · its alternatives →
16WM
We made glhf.chat – run almost any open-source LLM, including 405B▲161
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 · its alternatives →
17AT
A tiny LLM running at 21,000 tok/s on a $250 FPGA (Live Demo)▲79
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.
Aug 2026 · mikeayles.com · its alternatives →
18
Run Qwen3-Next-80B on 8GB GPU at 1tok/2s throughput▲123
Contribute to Mega4alik/ollm development by creating an account on GitHub.
2025 · github.com · its alternatives →
19
RunNburn – Run a 295B Moe from a 98GB GGUF on a 64GB RAM Desktop▲11
runNburn is an Apache-2.0 Rust inference engine for quantized GGUF models that are too big for your fast memory. The core idea: weights stay file-backed (mmap), host residency stays under an explicit byte budget (--ram-budget), and GPU caches are sized from detected free/total VRAM — never from device-name presets. There is no conversion step, no sidecar cache files, no silent requantization. The GGUF on disk is the single source of truth. The result that made me want to post this: Tencent's Hy3 (295B total / 21B active sparse MoE, a single 97.8 GiB Q2_K GGUF) runs on my desktop…
Jul 2026 · github.com · its alternatives →
20
Run 500B+ Parameter LLMs Locally on a Mac Mini▲17
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 · its alternatives →
21
A nibble-oriented CPU in Verilog to build a scientific calculator▲119
The core question: how did HP's scientific calculators actually work at the gate level? That rabbit hole led to building one from scratch. The architectural decision everything else follows from: a decimal calculator should store numbers as BCD — one decimal digit per 4-bit nibble. A standard byte-oriented CPU (Z80, 6502) fights that layout constantly. So I designed a small custom CPU in Verilog where 4 bits is the natural data width and memory is nibble addressable. What the project covers: - Custom CPU: Harvard architecture, 12-bit ISA, 8-state execution FSM, hardware stack guard with a…
May 2026 · github.com · its alternatives →
22
ZSE – Open-source LLM inference engine with 3.9s cold starts▲58
I've been building ZSE (Z Server Engine) for the past few weeks — an open-source LLM inference engine focused on two things nobody has fully solved together: memory efficiency and fast cold starts. The problem I was trying to solve: Running a 32B model normally requires ~64 GB VRAM. Most developers don't have that. And even when quantization helps with memory, cold starts with bitsandbytes NF4 take 2+ minutes on first load and 45–120 seconds on warm restarts — which kills serverless and autoscaling use cases. What ZSE does differently: Fits 32B in 19.3 GB VRAM (70% reduction vs FP16) — runs…
Feb 2026 · github.com · its alternatives →
23
Determinstic LLM inference for lowest price Gemma 4, with Windows XP▲9
Open-weight models, fully deterministic. The same answer every time, byte for byte, at the floor price.
Sep 2026 · tokendelivery.ai · its alternatives →
24
Cuts Long Horizon Inference Costs by 50% via external KV Cache Offload▲22
Hey HN, we’re the developers of OpenLake, an open source storage engine for offloading LLM KV caches from GPU memory into a shared tier of RAM and NVMe. We built OpenLake because KV caches are outgrowing GPU memory. A single 256K token conversation on Gemma 4 31B produces approximately 43GB of KV state, more than half the memory of an 80GB H100. The problem becomes even harder across a cluster: a prefix cached on one GPU host is unavailable when the next request lands on a different GPU, forcing the new GPU to repeat work the fleet has already completed. Once the KV cache is offloaded,…
Jul 2026 · github.com · its alternatives →

Also compare

Ranked by how close each launch is in meaning, then by votes. Prices were read from each product’s own site when checked and can change. Refine with your own description →