About ApX Machine Learning

LLM guides, model data, and engineer-grade reference guides, independently built.

The Mission

ApX Machine Learning is built for engineers who are actively building and deploying: developers who need answers (what fits their hardware, what it costs, which model to pick) and the understanding underneath those answers.

That means tools that give you numbers you can act on. As a natural extension of these tools, we maintain a small library of deep, open-access engineering courses to help you understand the math and mechanics underneath the systems you deploy.

What We Build

Tools

The VRAM Calculator answers the most practical questions in LLM deployment: will it run, and what will it cost.

Data

The LLM Directory, leaderboards, and trend data track every model that matters: specs, benchmarks, hardware requirements, and licensing.

Education

A collection of free, in-depth guides and a masterclass for engineers who want to understand the math and details past the tutorials.

Why This Exists

I originally built ApX Machine Learning as a set of AutoML (Automated Machine Learning) tools for my own personal use. As open-weights models like Llama and Qwen became viable to run locally, I started writing guides on how to run them and what GPUs they required.

The pivotal moment came with the launch of DeepSeek V3, when the industry realized open models could compete at the absolute top tier. At the time, ApX had the only comprehensive guide detailing the VRAM requirements and hardware constraints of running it, and traffic to the site surged.

Since then, ApX has transitioned from a small enthusiast resource into a professional hub helping developers deploy models at home and at a production scale. Today, it hosts the VRAM calculator, LLM directory, and reference guides used by thousands of engineers weekly.

Wei-Ming Thor

Founder

How the Data & Guides Are Built

Maintaining up-to-date hardware databases and engineering reference guides requires continuous research. To do this at scale without sacrificing technical accuracy, we use the same AI-assisted workflows we write about: specialized agents query academic papers and benchmark data, draft visualizations, and run verification checks.

The Goals

Cover what most courses won't. The most important problems in applied AI aren't covered in introductory courses. ApX covers them: fine-tuning on constrained hardware, serving models efficiently, evaluating what benchmark scores actually mean for your use case.

Keep serious content free. The highest-quality technical ML education sits behind expensive subscriptions or buried in papers that assume institutional access. ApX keeps its core content free, so engineers anywhere, regardless of budget, can go as far as they need to.

Independence

Rankings, benchmark results, and calculator output are never for sale. Ads are always labeled and never influence what the data says.

Start Here

The VRAM calculator and directory require no account. For extra depth on fine-tuning, inference optimization, or production deployment, feel free to check out the courses.