Not sure which model fits your Mac?
Use our free checker — select your hardware and get instant recommendations.
Pick the model before the setup
The LLMCheck index ranks the best model for each memory size: best local LLM by Mac RAM, 16 GB Macs, 24 and 32 GB Macs and coding models by RAM tier.
Best Local LLMs for MacBook Air (2026): 8, 16 and 24 GB
The best model for a fanless MacBook Air at 8, 16 and 24 GB, following the LLMCheck index’s RAM tiers, with estimated tok/s per Air and the thermal reality of long sessions.
Local LLMs on an Intel Mac: i7/i9 Tokens per Second (2026)
No Apple Silicon? You can still run local AI on an Intel Mac — here's what's realistic, the best small models, honest tok/s expectations, and when to upgrade.
How to Run Qwen on a Mac: Qwen3.8-27B, Qwen 3.6, Flash-Next
Run Qwen 3.6-35B-A3B, the index’s best pick from 24 GB (126 tok/s community-measured on an M5 Max), the dense Qwen3.8-27B, or Qwen3.8-Flash-Next on a 128 GB Mac.
Local AI Coding Assistant on Mac: Ollama + Continue (2026)
A private, offline coding assistant: Ollama serving Qwen 3.6-35B-A3B (North Mini Code on 24 GB, Mellum2.1 on 16 GB), wired into Continue, Zed or Cursor.
How to Use MCP (Model Context Protocol) with Local LLMs on Mac
Give your private, local model agentic tool access with MCP — run Qwen 3.6 in Ollama, add MCP servers (filesystem, web), and connect them. Fully offline.
How to Build a Local RAG System on Mac with Ollama
Build a private retrieval-augmented generation system on your Mac with Ollama, an embedding model and ChromaDB, with Python examples.
Function Calling & Tool Use with Local LLMs on Mac
Give a local model tools: function calling with Gemma 4, Qwen and Llama through Ollama, with Python and curl examples.
How to Fine-Tune a Local LLM on Mac with MLX (LoRA)
Fine-tune an open model on your own data, entirely on Apple Silicon, with MLX-LM and LoRA. Prepare a dataset, train, fuse, and run your custom model in Ollama.
How to Install Ollama on Mac (2026): Setup, MLX & Models
Install Ollama with the app, the install script or Homebrew, confirm the MLX runner (0.40+) and pull a first model sized to your memory.
Ollama MLX on Mac (0.40+): Check It, Speed, and Fixes
Ollama 0.40+ runs supported models on Apple’s MLX by default. Check which runner a model uses, what speed to expect, and fix the MLX error messages.
How to Run Llama 4 Locally on Mac — Scout & Maverick Guide
Hardware requirements, installation steps, and performance tips for running Meta's Llama 4 Scout on your Mac.
How to Install MLX on Mac: mlx-lm Setup & First Model (2026)
Install mlx-lm with pip or Homebrew, run a 4-bit model from mlx-community, use the Python API and benchmark plain decode.
LLM Quantization Explained: Q4, Q5, Q8 — Which Is Best for Mac?
Understand quantization levels, compare size vs quality tradeoffs, and pick the right quant for your Mac's RAM.
LM Studio Setup Guide for Mac — Download, Install & First Chat
Get LM Studio running on your Mac with a beautiful GUI. Download models, start chatting, and enable the local API.
Ollama Errors on Mac (2026): Exact Messages and Fixes
Every common Ollama error on a Mac by its exact message, from pull model manifest: file does not exist to mlx runner failed, with copy-paste fixes.
Local AI Troubleshooting Hub — Fix Common LLM Issues on Mac
Slow inference, a GPU that is not used, a model too large for memory, Ollama errors: the troubleshooting hub links every fix.
Why Is My Local LLM Slow on Mac? 7 Fixes for Faster Inference
Low tok/s on a Mac usually comes from a CPU fallback, memory pressure, the wrong quantization or a long context. Seven fixes, ranked by impact.
Model Too Large for Mac RAM: Solutions for Running Big LLMs Locally
What to do when a model will not load: smaller quantizations, mixture-of-experts alternatives and the macOS GPU memory limit.
Local LLM Not Using GPU on Mac? How to Enable Metal Acceleration
How to confirm a model is running on the Apple Silicon GPU through Metal, and how to fix it when it falls back to the CPU.
Frequently Asked Questions
What is the easiest way to run AI locally on Mac?
Install LM Studio, a free app with one-click model downloads and no terminal, or Ollama if you are happy to type one Terminal command. Both run on any Apple Silicon Mac, and since v0.40 Ollama runs supported models on Apple’s MLX engine by default.
Do I need to know how to code to run local AI on Mac?
No. LM Studio and Jan are complete graphical apps. Ollama needs a few Terminal commands, and the guides here give them ready to copy and paste.
Which guide should I start with?
Start with the LM Studio guide for a visual app, or the Ollama install guide if you are comfortable in Terminal. If something fails, the Ollama errors guide lists each exact error message with its fix.