Georgia Tech launched OMSCS in 2014 as an experiment with Udacity and AT&T, and it has since become the program every other online master's gets compared to. The program says it graduated 10,000 alumni in its first ten years. It searches at about 5,400 times a month in the US, more than any other program in this corner of the site, and it deserves the attention. The price is the headline. Georgia Tech's Bursar lists Fall 2026 OMSCS tuition at $227 per credit hour, plus an online learning fee of $212 a semester if you take under four credit hours and $531 if you take more.
The degree is 30 credit hours, ten three-credit courses. One course a semester works out to about $890 a term, so the whole thing lands between roughly $9,000 and $9,500 depending on how fast you go. Rates creep up a few dollars a year, but this is still cheap enough that many employers' tuition benefits cover most of it. What you are buying is the Georgia Tech MSCS. The transcript does not say online.
You pick a specialization, and two matter for readers of this site: Machine Learning and Artificial Intelligence (the latter used to be called Interactive Intelligence). There are four others: Computing Systems, Computational Perception and Robotics, Human-Computer Interaction and Computer Graphics. For ML, the courses people actually talk about are CS 7641 Machine Learning, CS 7643 Deep Learning, CS 7642 Reinforcement Learning and CS 6601 Artificial Intelligence. I have reviewed CS 7641 and CS 7643 separately on this site and both are hard, worthwhile courses. The weak points are real.
OMSCS runs at a scale where TAs do most of the teaching, and course quality varies. Some courses get refreshed regularly. Others are clearly running on lectures recorded years ago, and the lecture videos for a few ML courses feel dated next to what is now free on YouTube from Stanford and Karpathy. Graduate Algorithms, which the ML specialization requires, has a reputation as the course that stalls people near the end of the degree.
My advice is to read the OMSCentral reviews for every course before you register, because workload estimates there are more honest than the official pages. The other thing to be clear about is what the degree teaches. This is a computer science master's with an ML focus, not an LLM engineering course. You will come out understanding learning algorithms, optimisation and evaluation properly, which ages well. You will not come out with a polished agent framework portfolio unless you build one alongside it.
I think that is the right trade for most people who want to stay in ML for a decade, but it is the wrong trade if you need a job building RAG apps next quarter. My overall view: for a working programmer who wants a serious ML credential, OMSCS is the default answer, and the burden of proof is on any alternative that costs five times more.