Who it's for
For people who already use ChatGPT, Claude or Gemini and want to know what it takes to run or build on a model they can download. You need no machine learning background. A few terminal commands appear in lessons 1 and 4, and lesson 2 guides you through reading licenses without being legal advice.
Lessons
- 1What are open-weight models? Open weights vs open sourceWhat open-weight models are and how open weights differ from open source: inspect a real release on Hugging Face and meet the six model families.
- 2Open model licenses: what Apache 2.0, MIT, Llama and Mistral let you doOpen model licenses explained: what Apache 2.0, MIT, the Llama 4 license and Mistral's license let you do, and how to read one on a model card.
- 3How to choose an open model for your job and your computerHow to choose an open-weight model: match the job, read total and active parameters, check context length and license, then test two on your own prompts.
- 4Quantization explained: what Q4_K_M, GGUF, AWQ and GPTQ meanQuantization explained: read a GGUF name like Q4_K_M, see what a 4-bit file costs in accuracy, and compare GGUF, AWQ, GPTQ and bitsandbytes.
Frequently asked questions
Is the open-weight models course free?
Yes. Every lesson is free to read. The models they name have their own licenses, and lesson 2 shows how to read each one.
Do I need a GPU to follow the course?
No. The lessons read model cards, license files and documentation. Running a model takes enough memory for its file. The local hardware lesson has the sizing arithmetic, lesson 3 weighs total against active parameters, and lesson 4 covers the quantization that shrinks a file.
What should I know before I start?
Basic comfort with a terminal helps in lessons 1 and 4, which show a few commands. Nothing else is assumed.
Which models do the lessons use as examples?
gpt-oss, Qwen, DeepSeek, Llama, Gemma, Mistral and Olmo, as their cards read on 6 November 2026. New versions arrive every few months, so the lessons teach how to read a card and not a ranking.
What should I learn after this course?
The Local AI course is next. It starts with installing Ollama, then covers its API, LM Studio and the hardware a local model needs.