Best free courses · Generative AI and LLMs

Best free generative AI and LLM courses on YouTube, ranked by total hours

Updated 2026-10-13 · 19 courses · totals are live

Generative AI courses on YouTube come in two kinds. University courses, from Stanford, IIT Delhi and IIT Madras, explain how large language models work: attention, transformers, pretraining and alignment. Builder courses, from CampusX, Krish Naik and others, teach the tools for making things with them: LangChain, LangGraph, retrieval-augmented generation (RAG) and AI agents. Both kinds are long, and a playlist of nine videos can still be sixteen hours.

The numbers next to each course are live totals from the playlist itself, ranked longest first. Click one to see every video, tick off what you've watched, and get a finish date.

From 2h 26m (MCP (Model Context Protocol) — Krish Naik) to 2d 7h 37m (Stanford CS25: Transformers United — Stanford Online).

#	Course	Videos	Total	At 1.5×
1	Stanford CS25: Transformers United — Stanford Online Talks by researchers working on transformer models, one guest lecture per video. A series to dip into rather than a course to finish.	50	2d 7h 37m	1d 13h 5m
2	mini-RAG \| From notebooks to the PRODUCTION · mini-RAG: from notebooks to production — Abu Bakr Soliman A RAG application built step by step, from a notebook to a deployed service, taught in Arabic.	26	1d 9h 13m	22h 9m
3	Introduction to Large Language Models (LLMs) · Introduction to Large Language Models — NPTEL, IIT Delhi A full university course on large language models, from language modelling basics to transformers and training.	38	1d 5h	19h 20m
4	Stanford CS224N Natural Language Processing with Deep Learning I Spring 2024 I Professor Christopher Manning · Stanford CS224N: Natural Language Processing with Deep Learning (Spring 2024) — Christopher Manning The university course underneath today's language models: word vectors, neural networks, attention and transformers.	23	1d 3h 33m	18h 22m
5	Stanford CS336: Language Modeling from Scratch \| Spring 2026 · Stanford CS336: Language Modeling from Scratch (Spring 2026) — Stanford Online Building a language model end to end, from the tokenizer and the transformer to training, data and alignment.	18	1d 10m	16h 7m
6	Agentic AI using LangGraph · Agentic AI using LangGraph — CampusX (Hindi) AI agents with LangGraph, built up from the first graph to tools, memory and multi-step workflows.	28	23h 42m	15h 48m
7	Stanford CS236: Deep Generative Models I 2023 I Stefano Ermon · Stanford CS236: Deep Generative Models (2023) — Stefano Ermon The theory under image and text generation: autoregressive models, VAEs, GANs, normalizing flows and diffusion.	18	23h 38m	15h 45m
8	Generative AI using LangChain · Generative AI using LangChain — CampusX (Hindi) LangChain from models and prompts to chains, retrievers and RAG.	21	21h 53m	14h 35m
9	Neural Networks: Zero to Hero — Andrej Karpathy Neural networks coded from nothing in Python, ending with a GPT and its tokenizer built line by line.	10	19h 21m	12h 54m
10	Generative AI course — Sheryians AI School Five long sessions of more than three hours each on building with generative AI.	5	16h 53m	11h 15m
11	Stanford CME295: Transformers and Large Language Models I Autumn 2025 · Stanford CME295: Transformers and Large Language Models (Autumn 2025) How transformers and large language models work, in nine lectures.	9	16h 11m	10h 47m
12	LangChain and LangGraph Zero to Hero — LearningHub LangChain and LangGraph in one course, from the first chain to agents.	34	15h 3m	10h 2m
13	Introduction to Large Language Models — IIT Madras BS degree The large language models course from IIT Madras's online BS degree, in short lectures.	55	15h 3m	10h 2m
14	Agentic AI — Krish Naik Long hands-on sessions on building AI agents and agent workflows.	14	13h 38m	9h 5m
15	n8n with Karim — Kariim Nabiil Automating work with n8n, the workflow automation tool, taught in Arabic.	26	11h 13m	7h 28m
16	Agentic Coding using Claude Code — CampusX (Hindi) Building software with an AI coding agent, using Claude Code.	15	9h 51m	6h 34m
17	Agentic AI with LangGraph — Krish Naik A shorter route through LangGraph agents than CampusX's course.	12	9h 21m	6h 14m
18	RAG Playlist — CampusX (Hindi) Retrieval-augmented generation on its own: documents, embeddings, vector stores and retrieval.	10	8h 39m	5h 46m
19	MCP (Model Context Protocol) — Krish Naik Four sessions on MCP, the protocol for connecting AI models to tools and data. The shortest course here.	4	2h 26m	1h 38m

Totals count available videos only and update as the playlists change. Private or deleted videos are excluded. Compare the top 5 side by side →

How to choose

To understand how LLMs work: Karpathy's Zero to Hero if you learn by coding, Stanford CME295 or CS224N if you want lectures. CS336 is the deepest: you build the model yourself.
To build apps with them: a LangChain course, then a RAG course, then an agents course. CampusX's three follow that order in Hindi; Krish Naik's are shorter and in English.
New to machine learning? The machine learning courses come first, and every course here assumes Python.
These tools change fast. LangChain and LangGraph courses go out of date within a year, so prefer the newest version of a playlist; the university courses age more slowly.

Plan the course

Open a course here, set your speed and daily hours in the planner, and it gives you a finish date; tick videos as you go and the dates update. Lectures with code on screen don't speed-watch well, and the speed guide shows what 1.25× still saves. The guide to finishing a long course has the method.