Guide
How long does it take to learn machine learning on YouTube? 9 free courses, measured
Updated 2026-11-11 · 7 min read
A full free machine learning course on YouTube takes 26 to 63 hours of video. We measured the nine playlists on our machine learning list on 11 November 2026, counting every video to the last: Stanford's CS229 is 25 hours 43 minutes in its 2022 recording and 27 hours 51 in Andrew Ng's 2018 one, StatQuest's series 29 hours 50, Krish Naik's 36 hours 53 and CampusX's Hindi 100 Days of Machine Learning 62 hours 50.
The short ways in are much shorter. 3Blue1Brown's neural networks series is 3 hours 37 minutes, MIT's 2026 deep learning lectures 8 hours 19, and Stanford's gentler EE104 14 hours. At an hour a day, CS229 takes four weeks at normal speed, which is the speed its derivations need, and CampusX two months.
Every machine learning course, measured
Ranked by total. "1.5×" is the total watched at one and a half times speed, and "Avg" the average video.
| Course | Videos | Total | 1.5× | Avg |
|---|---|---|---|---|
| MIT 6.S191, all years | 90 | 73 h | 49 h | 49 m |
| CampusX | 134 | 63 h | 42 h | 28 m |
| Krish Naik | 153 | 37 h | 25 h | 14 m |
| StatQuest | 106 | 30 h | 20 h | 17 m |
| CS229, 2018 | 21 | 28 h | 19 h | 80 m |
| CS229, 2022 | 19 | 26 h | 17 h | 81 m |
| sentdex | 72 | 19 h | 13 h | 16 m |
| Stanford EE104 | 19 | 14 h | 9 h | 44 m |
| 3Blue1Brown | 10 | 4 h | 2 h | 22 m |
Totals are rounded to the hour; 3Blue1Brown's series is 3 hours 37 minutes exactly. CampusX teaches in Hindi, the other eight in English. None of the nine playlists has a private or deleted video. Each course name opens its live page, where the total is current and you can tick off what you have watched. The same courses are on our list of free machine learning courses, with notes on who each suits.
A playlist that holds nine years of one course
MIT's 6.S191 is re-recorded every spring, and its playlist keeps every year since 2018: 90 videos, 73 hours. One year's course is 8 to 10 hours (2023's 9 hours 36, 2024's 8 hours 15, 2025's 9 hours 43), and the newest, 2026, is the first nine videos: 8 hours 19 minutes, from neural networks through transformers and generative models to reinforcement learning. Watch one year, the newest, not the playlist.
Theory, practice, or a first look
The nine are three kinds of course, and the kind decides your month more than the total does.
University theory, 14 to 28 hours. CS229 is lectures of about 80 minutes in which the algorithms are derived on the board: linear models, generalisation, kernels, EM and reinforcement learning. The 2018 and 2022 recordings are the same course taught by different people, Andrew Ng and then Tengyu Ma and Chris Ré, so pick one. Both assume linear algebra, probability and some calculus. EE104, by Sanjay Lall and Stephen Boyd, is the gentler Stanford course: 19 lectures of 44 minutes, less proof, more intuition.
Practice with code, 19 to 63 hours. CampusX's 134 videos average 28 minutes and cover data cleaning, feature engineering and every classic model, in Hindi; its last videos, on imbalanced data and tuning with Optuna, are an hour each. Krish Naik's 153 videos are shorter, 14 minutes on average, and aimed at interviews. sentdex's 72, recorded in 2016 and 2017, write each algorithm from scratch in Python before using scikit-learn. StatQuest's 106 explain one algorithm or idea per video, 15 minutes in the middle, and work best beside one of the lecture courses rather than on their own.
A first look, 4 to 8 hours. 3Blue1Brown's ten videos show what a neural network is, how gradient descent and backpropagation work, and how transformers and language models work: 3 hours 37 minutes, the one to watch before anything else. MIT's 2026 lectures are the fastest tour of current deep learning, at 8 hours 19.
Days to finish at an hour a day
How many days each course takes at an hour a day, at normal speed and at 1.5×.
| Course | Hours | 1 h a day at 1× | 1 h a day at 1.5× |
|---|---|---|---|
| 3Blue1Brown | 3½ | 4 days | 3 days |
| MIT 6.S191, 2026 | 8 | 9 days | 6 days |
| Stanford EE104 | 14 | 15 days | 10 days |
| sentdex | 19 | 19 days | 13 days |
| CS229, 2022 | 26 | 26 days | 18 days |
| CS229, 2018 | 28 | 28 days | 19 days |
| StatQuest | 30 | 30 days | 20 days |
| Krish Naik | 37 | 37 days | 25 days |
| CampusX | 63 | 63 days | 42 days |
These are days of video. Theory lectures don't speed-watch well: a derivation at 1.5× is one you will rewind, so plan CS229 and EE104 at 1× (what 1.5× and 2× cost →). The practical courses go faster, but add the time to run the code yourself, which is where it sticks.
Put together, a theory route is 3Blue1Brown, then CS229 2018 with StatQuest open for whenever a lecture loses you: 31½ hours of lectures, about a month at an hour a day at 1×. A practical route is CampusX or Krish Naik, then MIT's 2026 lectures for deep learning: 45 to 71 hours.
Check the total covers every video
Some playlist length tools add up only the first 100 videos and stop without saying so. Three of these playlists are longer than that, and the part a capped tool misses is not small: on CampusX's 134 videos it is the last 34, 17 hours 53 minutes, more than a quarter of the course; on Krish Naik's 153 it is 53 videos, 12 hours 5; on StatQuest's 106 it is six videos on reinforcement learning and RLHF, 2 hours 28. If a total looks short for a playlist with more than 100 videos, here is how to check it.
Plan the one you choose
Paste the course's link on the home page, or click its name in the tables above, and set your daily minutes in the planner: it gives you a finish date and the videos for each day, and you can put the sessions in your calendar. Tick each video as you finish it and the time left updates; the course stays in your list on the home page, so you can pick it up where you stopped. Machine learning usually comes after Python: the Python guide measures those courses the same way, and how long a free YouTube course is compares the subjects.
Frequently asked
How long does it take to learn machine learning?
A full free course on YouTube is 26 to 63 hours of video, measured on 11 November 2026: Stanford's CS229 26 to 28, StatQuest 30, Krish Naik 37, CampusX 63. At an hour a day that is one to two months, before the time to practise. An introduction takes an afternoon: 3Blue1Brown's series is 3 hours 37 minutes.
How long is Andrew Ng's CS229?
The Autumn 2018 recording is 27 hours 51 minutes over 21 videos of about 80 minutes. At an hour a day at normal speed, which suits its derivations, that is 28 days. The Spring 2022 recording of the same course is 25 hours 43.
How long is CampusX's 100 Days of Machine Learning?
62 hours 50 minutes over 134 videos, in Hindi, so more than the hundred its name suggests. At an hour a day it takes 63 days at normal speed and 42 at 1.5×.
Can I learn machine learning in a month from YouTube?
You can watch one course in a month: CS229 at an hour a day, or Krish Naik's at 1.5×. Understanding it takes the maths underneath and practice with real data, so a month is the lectures, not the skill.
Which free machine learning course is the shortest?
3Blue1Brown's neural networks series, 3 hours 37 minutes over ten videos. It explains neural networks rather than all of machine learning; for a short full course, Stanford's EE104 is 14 hours.
Why might a total differ from the playlist today?
These totals were measured on 11 November 2026. Creators add videos, MIT adds a new year each spring and StatQuest still adds to its series, and videos go private. Each course's page here, and the machine learning list, shows the total as it is now.