The verdict in 60 seconds
Columbia does not give away degrees, but you can study a surprising amount of its real curriculum for nothing. The best free route in 2026 is auditing ColumbiaX courses on edX — including all four graduate-level courses of the Artificial Intelligence MicroMasters — plus Columbia-taught Coursera courses such as Perry Mehrling’s Economics of Money and Banking. Auditing gets you the lectures and most reading material. It does not get you graded assignments or a certificate. If you need something to show an employer, budget roughly £40–£250 per course for a verified certificate on edX, or about £45 a month on Coursera. If you only want the knowledge, you can get a serious Ivy League education in AI, finance, and economics without spending a pound.
Columbia University runs its online outreach through two channels: ColumbiaX on edX, where its most technical material lives, and a smaller set of long-running Coursera courses taught by Columbia and Barnard faculty. The problem with most “free Columbia courses” articles is that they blur two different things — courses that are free to audit, and courses that hand you a free certificate. Almost nothing from Columbia falls into the second category. What you actually get for free is access: the same lecture videos, and usually the same readings, that paying learners see. Whether that is enough depends on why you are studying. This guide lists what is genuinely available free in 2026, what each course demands of you, what the paid upgrade costs in pounds, and — because nobody else seems willing to say it — where each course falls short. No affiliate spin, just what holds up.
How we chose these courses
Every course below had to pass four tests. First, it must be genuinely free to audit in July 2026 — not a seven-day trial, not a “first module free” teaser. Second, it must be taught by Columbia or Barnard faculty, not licensed third-party content wearing a Columbia badge. Third, it must be substantial: multi-week, university-level material with real assessments on the paid track, so auditing still gives you a proper course rather than a webinar. Fourth, it must be current — we checked each course was open for enrolment this month, since edX quietly retires older runs. Prices are converted to pounds and were checked in July 2026; edX and Coursera both adjust pricing by region, so treat every figure as approximate.
The best free Columbia courses in 2026
1. Artificial Intelligence MicroMasters (edX)
Columbia’s flagship online offering: four graduate-level courses — Artificial Intelligence, Machine Learning, Robotics, and Animation and CGI Motion — drawn from its computer science masters programme. Each course runs 8–12 weeks at 8–10 hours a week, so the full sequence is close to a year of part-time study. Audit is free; the verified track costs roughly £200 per course, and the full MicroMasters credential can count toward Columbia’s Computer Science MS if you are later admitted. Who it is for: people with solid maths (linear algebra, probability, calculus) and working Python. The weak point: auditors get little forum support, and the assignments — the best part — sit behind the paywall.
2. Machine Learning — ColumbiaX (edX)
The strongest single course inside the MicroMasters, taught by John Paisley. Twelve weeks, 8–10 hours a week, covering machine learning from a probabilistic angle: maximum likelihood, Bayesian methods, Gaussian processes, hidden Markov models. It is what you take after an introductory ML course, when you want to understand why the algorithms work rather than just call them from scikit-learn. Free to audit; verified certificate roughly £200. Be honest with yourself about prerequisites — learners without real linear algebra and probability tend to stall around week four. If that is you, start with Essential Math for AI below and come back.
3. Corporate Finance Professional Certificate (edX)
Three short courses from Columbia Business School — Introduction to Corporate Finance, Free Cash Flow Analysis, and Risk & Return — taught by finance faculty including Daniel Wolfenzon. Each runs about four to six weeks at 3–5 hours a week, and each can be audited free on its own. You learn discounted cash flow valuation, how to read what a company’s cash flows are actually telling you, and how risk gets priced. Who it is for: analysts, founders, and career-changers who want finance fundamentals from a top business school without an MBA price tag. Verified certificates run roughly £50–£75 per course. The limitation: it is lecture-driven, and without the paid track’s assessments you will need discipline to practise the valuation work yourself.
4. Financial Engineering and Risk Management Specialization (Coursera)
A five-course sequence from Columbia’s Industrial Engineering and Operations Research department, built by faculty including Garud Iyengar and Martin Haugh. This is proper quantitative finance: derivatives pricing, term-structure models, portfolio optimisation, and computational methods. Each course can be audited free individually; the certificate route needs a Coursera subscription at about £45 a month. Time commitment is real — expect five to six months at 4–6 hours a week for the full sequence. Who it is for: maths or engineering graduates aiming at quant and risk roles. Who it is not for: anyone hoping for a gentle “how markets work” introduction. The exercises assume comfort with calculus and probability from day one.
5. Economics of Money and Banking (Coursera)
Perry Mehrling’s cult classic from Barnard College, Columbia. Thirteen weeks at 4–6 hours a week on how the monetary system actually works: dealers, repo markets, central bank balance sheets, and the “money view” of the economy. There is nothing else like it online, free or paid, and professional economists still recommend it a decade after it first ran. Completely free to audit, with a certificate available for roughly £38 if you want one. Who it is for: anyone serious about finance, central banking, or economic journalism. The honest weakness: the lectures were recorded years ago, so recent events are absent — though the framework is exactly what you need to interpret them.
6. The Age of Sustainable Development (Coursera)
Jeffrey Sachs’ broad survey of sustainable development: economic growth, planetary boundaries, health, education, and the UN Sustainable Development Goals he helped design. Fourteen weeks at a light 2–3 hours a week, free to audit, certificate around £38. It is the most accessible course on this list — no maths, no prerequisites — and works well as a structured alternative to reading a stack of policy books. Who it is for: students weighing a development or policy career, and NGO or public-sector workers wanting the big picture. The weakness is age and angle: parts of the data are dated, and Sachs’ own policy positions are presented more confidently than some economists would accept.
7. Essential Math for AI — ColumbiaX (edX)
A newer ColumbiaX course built for exactly the gap that sinks most self-taught ML learners: the mathematics. Linear algebra, calculus, probability, and optimisation, taught with AI applications in view rather than as abstract theory. Self-paced, roughly six to eight weeks at 4–5 hours a week. Free to audit, with a verified certificate in the region of £150. Who it is for: anyone planning to attempt the AI MicroMasters, or any serious ML study, without a recent maths background. It is drier than the AI courses it prepares you for — that is the nature of the material — but working through it first will save you from abandoning Machine Learning at the midpoint like so many auditors do.
8. Robotics — ColumbiaX (edX)
Matei Ciocarlie’s graduate robotics course, the third leg of the AI MicroMasters. Twelve weeks at 8–10 hours a week covering robot kinematics, motion planning, and manipulation, with project work built around simulated robot arms in Python. Free to audit; verified track roughly £200. Who it is for: engineers and CS graduates curious whether robotics is their field — it is one of very few university-level robotics courses available free anywhere. Two honest caveats: the projects, which are the heart of the course, are only assessed on the paid track, and the Python-plus-linear-algebra prerequisite is enforced by the material itself within the first fortnight.
Common mistakes to avoid
Four pitfalls catch most people who start down the free-Columbia route.
1. Confusing “free to audit” with “free certificate.” Columbia offers no free certificates. If a job application needs proof, price the verified track before you invest ten weeks — retrofitting a certificate after finishing an audit is sometimes impossible on older course runs.
2. Starting the AI MicroMasters without the maths. The single most common failure pattern. These are genuine graduate courses; enthusiasm does not substitute for linear algebra. Take Essential Math for AI first if in doubt.
3. Paying a Coursera subscription up front, then stalling. At about £45 a month, a six-month drift costs £270. Audit free until you are certain you will finish, then subscribe for the final stretch and complete the graded work in a month or two.
4. Treating certificates as job tickets. UK employers read a ColumbiaX certificate as evidence of self-directed learning, not as a Columbia credential. What moves interviews is what you built or can explain — so keep the projects, notes, and code, not just the PDF.
Frequently asked questions
Are Columbia’s free online courses really free?
Yes, in the audit sense. On edX and Coursera you can watch every lecture and read most materials in the courses above without paying. What is excluded: graded assignments, instructor feedback, and certificates. On edX, audit access can also be time-limited to the course run, so do not enrol and then leave it for six months.
Do I get a certificate from a free Columbia course?
No. Certificates always cost money: roughly £38–£49 per course on Coursera, and roughly £40–£250 on edX depending on the course. edX runs a financial assistance scheme that can cut verified-track fees by up to 90% if you apply and qualify — worth doing if the price is the only barrier.
Is the AI MicroMasters worth paying for?
Pay only if one of two things is true: you want the graded projects and exams to force depth, or you plan to apply to Columbia’s CS masters, where the credential can count toward the degree. If you simply want to learn the material, audit free and build your own projects — the lectures are identical either way.
Can free courses lead to an actual Columbia degree?
Only indirectly. The AI MicroMasters is the one pathway: complete the paid verified track and, if admitted to Columbia’s Computer Science MS, it can count toward the degree. Auditing alone carries no credit anywhere. No Coursera course from Columbia currently stacks into a degree programme.
How do these compare with non-university alternatives?
Columbia’s courses are deeper and slower. If you want employable skills fast, freeCodeCamp or a well-chosen Udemy course will get you building sooner. Columbia wins when you need theory that lasts — the maths behind ML, or how money markets really work. Many learners sensibly do both: practical platform first, Columbia for depth.
Related guides
If you are weighing platforms rather than universities, start with our full Coursera review, which covers how auditing and subscriptions actually work. For cheaper practical alternatives, see our Udemy guide and our freeCodeCamp review for 2026 — for coding skills specifically, the free option is often the better one.