Back to index
OtherScheduled and self paced courses from a few weeks to a full semester, plus MicroMasters programmes of four to six courses·Courses are free to take; certificates run about $49 to $300 per course, and MicroMasters bundles range from $1,260 to about $2,965

MITx on MIT Learn (Platform Review)

4.4

MITx started the whole MOOC era in 2012 and has spent the years since the edX sale taking its courses back. MIT Learn is the result: one catalogue for MITx courses, MicroMasters programmes and OpenCourseWare materials, with an AI assistant on top. The content is the best there is at this price and the migration is still untidy, so check where your course actually lives before enrolling.

What We Liked

  • Every MITx course can be taken free, with certificates priced sensibly at roughly $49 to $300 rather than the $299 edX now charges for comparable courses
  • 226 courses and 20 programmes in the MITx unit alone, including Introduction to Machine Learning, the probability and statistics sequence and four MicroMasters credentials
  • MIT Learn is run by MIT Open Learning, a university, not by 2U, the restructured owner of edX
  • AskTIM, the platform's AI assistant, is useful for finding courses and summarising syllabi across 12,700 resources
  • OpenCourseWare lecture notes and videos sit in the same search, so you can pair a graded course with the full residential materials

What Could Be Better

  • Three generations of platform (edX, MITx Online, MIT Learn) still coexist, enrolments are being migrated, and the old mitxonline.mit.edu address now simply redirects
  • Search mixes free learning materials with graded courses and paid programmes, so results need filtering before they make sense
  • The flagship MITx courses run on fixed calendars with hard deadlines, which is right for rigour and wrong for people who need true self pacing
  • Most of the AI-adjacent catalogue is foundations: there is little on applied generative AI or current tooling
  • Pricing is inconsistent across listings, with some certificates showing ranges and some programmes quoted to the cent

Detailed review

A quick history, because it explains the mess. MIT launched MITx in December 2011 and ran its first course, 6.002x Circuits and Electronics, in spring 2012. Two months later MIT and Harvard founded edX as a nonprofit to host MITx and HarvardX courses. In 2021 the two universities sold edX to 2U for about $800 million, and MIT's Office of Digital Learning built MITx Online on the open source Open edX code to keep running its own courses.

2U filed for Chapter 11 in July 2024 and emerged as a private company owned by its lenders. Then in July 2025 MIT launched MIT Learn, a single platform for all of its non-degree learning, and MITx Online was absorbed into it. Today mitxonline.mit.edu redirects to learn.mit.edu, and MIT's support pages say many MITx courses are now offered exclusively on MIT Learn, with edX enrolments being moved over. What you get is a catalogue. The MITx unit on MIT Learn lists 459 results: 226 courses, 20 programmes and 213 learning materials, with 41 items carrying a MicroMasters credential.

Across all of MIT Learn there are more than 12,700 resources from MITx, OpenCourseWare, xPRO, Professional Education and Sloan Executive Education, and MIT says 98 percent of them are free. For readers of this site the relevant corner is the statistics and machine learning sequence: 6.431x Probability, 18.6501x Fundamentals of Statistics, 6.86x Machine Learning with Python, 6.419x Data Analysis, 14.310x Data Analysis in Social Science, Introduction to Machine Learning, and the MicroMasters in Statistics and Data Science that bundles four of them for $1,350. There are also the Supply Chain Management, Finance and Principles of Manufacturing MicroMasters at $1,525, $2,965 and $1,260 respectively. Pricing is the pleasant surprise. Single-course certificates I checked ran from $49 for Just Money and Product and Service Creation in the Internet Age to $99 for u-lab, $149 for the user innovation course, and $300 for each course in the statistics and data science sequence.

edX now charges $299 for a certificate on comparable Harvard courses. MIT keeping its certificates under its own roof and under $300 is one of the clearer benefits of the move. AskTIM is the platform's AI layer. It recommends courses from a goal, summarises a course's structure and expectations, and in selected courses acts as a syllabus chatbot and tutor. It is useful for navigation in a catalogue this size, and MIT's framing of it as an initial launch is honest.

A chatbot is not a teaching assistant, and the forums in the serious courses are still where help actually happens. The problems are the ones you would expect from a migration. Three generations of platform coexist, and if you enrolled in an MITx course on edX in the past few years you may find your enrolment, your progress and your verified identity in different places. That matters most for the MicroMasters, where the capstone exam requires the same verified identity across every course. MIT's support centre says learners will receive instructions before a course transitions, and from autumn 2026 the SDS courses run only on MIT Learn, so the instruction is clear enough: enrol on MIT Learn and nowhere else.

Search is the other irritant. Learning materials, which are OpenCourseWare notes and videos with no grading, sit beside graded courses and paid programmes in the same results, so the filters do real work. And the pricing display is inconsistent, with one social science course showing a certificate range of $250 to $1,000 and the programmes quoted to the cent. On content, MITx is foundations. Probability, statistics, calculus, circuits, thermodynamics, machine learning from first principles.

If you want applied generative AI, prompt engineering or current tooling, this is the wrong catalogue, and I think that is fine. There are a hundred places to learn the tools and a handful to learn why they work. My own take is that MIT Learn is the right answer to a bad situation. The university that invented the MOOC sold its platform, watched the buyer go bankrupt, and took its courses home. The home is still being built, and for the next year you should expect redirects and migration emails.

The courses themselves remain the standard against which I judge everything else on this site, and they are free unless you want the paper.

[ final ]

The verdict.

If you want MIT courses, go to MIT Learn rather than edX from now on, audit for free, and pay for a certificate only where a credential matters. The platform is better than it was and messier than it will be. The teaching has never been the problem.