The verdict in 60 seconds
Yes, you can take real Stanford courses for free in 2026, but “free” comes in two different flavours and it matters which one you pick. If you want a certificate and a structured schedule, look at the Stanford-affiliated courses on Coursera and edX (Machine Learning Specialization, Statistical Learning) — you can audit the video content for £0, but the certificate usually costs £39-£79 or a Coursera Plus subscription at roughly £44/month. If you just want to learn and don’t care about a credential, Stanford also publishes full lecture courses (CS229, CS231n, CS224n) on YouTube with zero paywall at all, and runs Code in Place, a genuinely free, live, instructor-led Python course every year. Our pick for most beginners: start with Code in Place if it’s enrolling, or CS229 on YouTube if you want to start today.
Stanford doesn’t have a single “free courses” portal that lists everything in one place, which is part of why this is confusing. The university spreads its free content across four different channels: Stanford Online’s own site, Coursera (where Andrew Ng’s machine learning courses live), edX (where the statistics and database courses live), and a public GitHub/YouTube layer that most people never find because it isn’t advertised on the main Stanford Online homepage. Some of it is genuinely free with zero catches. Some of it is “free to audit, pay if you want proof you did it.” A few pages that used to host free Stanford MOOCs (the old Lagunita platform) were retired in 2020 and folded into edX, so if you’re finding old blog posts pointing to lagunita.stanford.edu, expect a redirect or a dead link.
Below is what’s actually live and enrollable in 2026, sorted roughly by how useful they are for someone trying to build real, job-relevant skills rather than just collect a Stanford name on a certificate. We’ve separated “structured platform courses” from “raw lecture archives” because they solve different problems: the former are better if you need accountability and a certificate for LinkedIn, the latter are better if you already know how to learn independently and just want Stanford-quality material for nothing.
How we picked these
We only included courses that meet four criteria. First, the course has to be currently live and enrollable as of 2026 — we checked each link and excluded anything redirecting to a dead Lagunita page or an archived, non-enrollable course shell. Second, it has to be taught or created by Stanford faculty, not just “Stanford-adjacent” content from a former student or an unaffiliated bootcamp using the Stanford name loosely. Third, the free tier has to be real: full video content and materials at £0, not a “free trial” that locks after seven days. Fourth, we prioritised courses with a clear, specific learning outcome over generic “introduction to X” survey courses, because those tend to be the ones people actually finish.
1. Code in Place (Stanford CS106A)
Code in Place is the closest thing to actually attending Stanford for free. It’s the online version of CS106A, Stanford’s largest on-campus course, condensed to its first six weeks and taught live by volunteer section leaders (many of them CS106A alumni) in small groups over video call. In its most recent run the program had roughly 1,000 volunteer instructors teaching around 17,000 students from more than 180 countries. This is a genuinely different experience from a pre-recorded MOOC: you get live sessions, homework feedback, and a small cohort of classmates. It’s aimed squarely at complete beginners — no programming background assumed — and covers Python fundamentals including control flow, loops, console programs, basic graphics, and lists and dictionaries. There’s no cost at any point and no upsell to a paid certificate; participants who finish get a certificate of completion from Stanford, not course credit. The catch is timing: it runs in specific windows each year (historically spring), so you need to catch an open enrolment period rather than starting whenever you feel like it.
2. Machine Learning Specialization (Andrew Ng / DeepLearning.AI + Stanford Online)
This is the updated, three-course version of the original Andrew Ng Machine Learning course that has now been taken by more than 4.8 million learners since 2012, produced jointly by DeepLearning.AI and Stanford Online and hosted on Coursera. It covers supervised learning (regression and classification), plus an introduction to neural networks, decision trees, and unsupervised learning, all built around practical Python exercises rather than heavy proofs. You can audit every video and reading for free. The catch is the graded assignments and the certificate: Coursera gates those behind either a one-time payment (roughly £39-£79 depending on the course) or a Coursera Plus subscription (around £44/month, with an annual plan usually working out cheaper if you’ll use it for other courses too). Budget 3-4 months at roughly 5 hours a week if you’re doing all three courses properly with the assignments. It’s a genuinely good on-ramp to ML for people who want the maths explained simply rather than rigorously — for the rigorous version, see CS229 below.
3. Statistical Learning with R or Python (Hastie & Tibshirani)
Taught by Trevor Hastie and Robert Tibshirani, two of the statisticians who wrote the field’s standard textbook, this course is available on Stanford Online and edX in separate R and Python versions. It closely follows their book “An Introduction to Statistical Learning” (Springer), which is legally free to download as a PDF from the authors’ own site — so even if you don’t take the course, the reading material costs nothing. The course covers linear and logistic regression, cross-validation, regularisation (ridge and lasso), tree-based methods, random forests, boosting, support vector machines, and a brief introduction to neural networks and deep learning. You can audit the full course free on edX; a verified certificate costs extra (edX typically prices these in the £120-£160 range, though pricing varies by run). This is a stronger statistical foundation than the Andrew Ng specialization above, and better suited to people who already have some linear algebra and want to understand why the models work, not just how to call them in code.
4. CS229: Machine Learning (lecture archive)
CS229 is Stanford’s actual graduate-level machine learning course, and full lecture recordings are publicly posted on YouTube for free, alongside the course notes and problem sets that Stanford makes publicly available on the course website. There’s no certificate track and no grading, so this only works if you’re self-motivated and comfortable with multivariable calculus, linear algebra, and probability going in — it moves considerably faster and deeper than the Coursera version. What you get in return is the real Stanford graduate curriculum with no paywall anywhere: derivations of gradient descent, the EM algorithm, support vector machines, and the theoretical underpinnings that the more beginner-friendly courses skip. Time commitment is roughly what you’d expect from a full graduate course — plan for several months if you work through the problem sets seriously, or a few weeks if you’re just watching lectures for understanding.
5. CS231n: Deep Learning for Computer Vision
This is Stanford’s flagship computer vision course, and past years’ lecture videos are freely available on YouTube, with the current course notes, slides, and assignments hosted openly on the CS231n GitHub Pages site. It covers convolutional neural networks from first principles through to modern architectures, image classification, object detection, and generative models. It assumes you already know Python and basic machine learning (CS229 or equivalent is good preparation), so this isn’t a starting point — it’s a strong second or third course. Everything is free with no certificate offered, which keeps this firmly in “learn because you want to know” territory rather than “collect a credential” territory. If you’re building a portfolio for a computer vision role, working through the assignments (which are publicly posted, unlike some other universities’ course materials) is worth more than most paid alternatives.
6. CS224n: Natural Language Processing with Deep Learning
The NLP counterpart to CS231n, covering word vectors, recurrent networks, attention, and transformer architectures — genuinely current material given how much NLP has changed since transformers took over. Full lecture videos are posted free on YouTube each year the course runs, and lecture notes and assignments are public on the course website. Same trade-off as CS231n: no certificate, no hand-holding, and you need solid Python and some prior deep learning exposure. Given that transformers now underpin most of the AI tools people use daily, this is arguably one of the most practically relevant free courses on this entire list if you want to actually understand what’s happening under the hood of large language models rather than just use them.
7. Databases (Jennifer Widom)
Originally one of Stanford’s first three MOOCs back in 2011, this course was rebuilt as a set of self-paced mini-courses covering relational databases, SQL, XML, NoSQL, and database design, and is now hosted on edX after the old Lagunita platform was retired in 2020. You can mix and match the mini-courses depending on what you actually need — someone who just wants solid SQL doesn’t have to sit through the XML modules. Audit access is free; edX charges for a verified certificate if you want one (again, typically £120-£160 depending on the specific mini-course and current promotions). For anyone working toward a data analyst or data engineer role, the SQL fundamentals mini-course alone is worth doing even without paying for the certificate — the material is unusually rigorous compared to typical “SQL in a weekend” tutorials.
8. Stanford Online’s free content library and YouTube channel
Beyond individual courses, Stanford Online maintains a dedicated “Free and Low-Cost Courses” section on its own site plus a general content library of talks, webinars, and short lecture series spanning health and medicine, engineering, education, and the arts and humanities — and a YouTube channel with hundreds of recorded talks and full course series that never had a paid tier to begin with. This is the least structured item on the list and the easiest to overlook, since Stanford doesn’t heavily promote it outside its own site. It’s worth a browse if your interest is broader than computer science — there’s real content here in child nutrition, health policy, and humanities that doesn’t fit neatly into a “top Stanford courses” listicle but is still legitimate Stanford faculty material, free, no account required beyond a YouTube login.
Common mistakes people make chasing “free Stanford courses”
- Assuming “free” always means a free certificate too. On Coursera and edX, free usually means free to audit the videos and readings. The certificate — the thing that actually shows up on LinkedIn — is a separate paid add-on in almost every case. Read the enrolment page carefully before assuming you’ll walk away with a credential.
- Trying to start with CS229 or CS231n with no background. These are Stanford’s actual graduate courses, not “for everyone” MOOCs. If you don’t already have linear algebra, calculus, and solid Python, you’ll bounce off in week one and conclude Stanford courses are “too hard,” when really you picked the wrong entry point. Start with Code in Place or the Machine Learning Specialization instead.
- Following old links to Lagunita. Plenty of “best free Stanford courses” articles online are three to five years old and link straight to lagunita.stanford.edu, which was shut down in 2020. If a link doesn’t redirect cleanly to edx.org or online.stanford.edu, assume the course has moved or been retired and search for the current version by name.
- Missing the Code in Place enrolment window. Unlike the Coursera and edX courses, which are self-paced and open year-round, Code in Place runs on a fixed annual schedule. People bookmark it, forget about it, and then find enrolment closed for months. If you want it, check the site or subscribe to updates rather than waiting until you’re ready to start.
Frequently asked questions
Are these free Stanford courses actually taught by Stanford professors?
Yes, in every case on this list. Code in Place, CS229, CS231n, and CS224n are Stanford courses taught by Stanford faculty and current or former teaching staff, made public through Stanford’s own channels. The Machine Learning Specialization, Statistical Learning, and Databases courses are built and taught by Stanford professors (Andrew Ng, Trevor Hastie, Robert Tibshirani, and Jennifer Widom respectively) but distributed through Coursera and edX rather than Stanford’s own site, which is why some people assume they’re not “real” Stanford courses. They are — the platform is just a distribution partner.
Do I get actual Stanford credit for finishing any of these?
No. None of the free courses on this list carry Stanford academic credit, even the ones that cost money for a certificate. A Coursera or edX certificate is a record of course completion from that platform (co-branded with Stanford where relevant), not a transcript entry or degree credit. If you need actual Stanford credit, that requires separate paid enrolment through Stanford’s professional education or graduate programs, which is a completely different (and far more expensive) track.
Which one should a complete beginner start with?
If Code in Place is enrolling, start there — it’s built for people with zero programming experience and gives you live human instructors, which nothing else on this list offers. If it’s between enrolment windows, start with the Machine Learning Specialization’s first course, which is also built for beginners and available year-round. Don’t start with CS229, CS231n, or CS224n unless you already have the maths and programming background — they will not hold your hand.
Is it worth paying for the certificate on Coursera or edX?
It depends what you’re using it for. If you’re job-hunting and need something concrete to show on LinkedIn or a CV, a certificate from a named Stanford-affiliated course carries some weight, particularly for the Machine Learning Specialization, which is widely recognised by hiring managers in tech. If you’re just learning for yourself, or you already have a portfolio of projects that demonstrates the skill, the certificate adds little — the free audit gives you the same knowledge for £0.
How much time should I realistically budget?
Code in Place is a fixed six-week commitment with live sessions, so budget 3-5 hours a week for that window specifically. The Machine Learning Specialization is roughly 3-4 months at 5 hours a week if you do all three courses with assignments. Statistical Learning and Databases are more flexible self-paced courses you can stretch or compress — most people finish either in 6-10 weeks at a few hours a week. CS229, CS231n, and CS224n depend entirely on how deep you go: watching lectures for understanding might take a few weeks, but working through the actual problem sets properly is a multi-month project.