OpenAI Built You a Free AI Degree Too: All 14 OpenAI Academy Courses, Reviewed

If you start on Monday, November 9, you have 53 days left in 2026. That is enough time to finish a whole AI curriculum from OpenAI, the company behind ChatGPT: fourteen courses, four tracks, a badge for every course and a certificate when you finish a full track. All of it free.

Here is the catch. Almost 70,000 people signed up for the first course. Fewer than 19,000 signed up for the third one, the one that finishes the certificate. So we went through all fourteen and gave each one a verdict: take it, take it if (meaning only for a certain person), or skip it. Then we built an eight-week plan that gets you up to three certificates before 2027.

OpenAI built one too: from 3 courses to 14

A few weeks ago we reviewed Anthropic's free AI degree. This is the other company everyone knows. Back in August, OpenAI Academy had three courses with a certificate, and we earned one on camera. Since then it has grown to fourteen courses, with a badge you can share for each one.

They are built for different people. Some are for anyone with a job. Some are for people who write code. One is for managers. Two are for teachers and students. So rule number one: don't start at the top and take them all.

Badge vs certificate, and the sentence most people miss

Every course ends with a short quiz: ten to twenty questions, pulled from a bank of fifty. Score 80% or more and you get a badge. Fail, and you just try again with new questions. You can also take a course without taking the quiz at all.

Finish every course in a track and you also get a certificate of completion. But only three tracks give you one: Foundations, Codex and the API track. Leadership, Educators and College Students give you a badge, but no certificate.

Now, the sentence. It is right on OpenAI's help page: "These badges and certificates of completion are not certifications and do not guarantee eligibility for a future certification."

Under our last video, one viewer called this kind of thing a useless badge. They are half right. It is not a degree, and it is not an exam that proves you are an expert. What it is, is proof you finished real practice, from the company that makes the tool. And most courses have you work on a real task, so you walk out with something you can actually show.

All 14 courses at a glance

Times are OpenAI's own estimates, before the quiz.

Track 1: Apply AI at work (Foundations certificate)

Three courses, and this is the track almost everyone should finish.

AI Foundations

About an hour. How large language models (the kind of AI behind ChatGPT) work, how to give clear instructions, how to add the right context, and how to check what comes back. About 70,000 people signed up for it, and it is the right place to start. Take it.

Applied AI Foundations

About an hour and a half. This is where it gets useful. You pick a task you do every week, break it into steps, decide where ChatGPT helps, and add points where you check its work. You leave with a better version of your own task. Take it.

Agents and Workflows

Another hour and a half. Instead of asking ChatGPT one question at a time, you hand it a whole job. You give it context, set limits, review the draft and reuse what works. One thing to know: the course asks for access to ChatGPT's agent features, so check your plan before you start. Take it. Finish this one, and that is your first certificate.

Track 2: Build with AI, the 5 API courses

Eight courses, split in two. Five are for building apps with the OpenAI API (the way programmers plug OpenAI's models into their own software), and three are for Codex, OpenAI's coding agent. These are for people who already write code, and every one ends with a hands-on lab.

• Scope AI Solutions, 30 minutes. How to find a problem worth solving with AI and turn it into a plan. Short, and useful for product managers too. Take it if you build or plan products.
• Evaluate AI Applications, 70 minutes. The best course in the builder track, and one of the least popular: about 1,500 people signed up. It teaches you to test whether your AI app is actually getting better, instead of just hoping. Take it if you build.
• Design and Build Agentic Systems, almost two hours, the longest in this track. Several agents, each with one clear job, limits on which tools they can use, and clean handoffs between them. Take it if you build agents.
• Build with Retrieval-Augmented Generation, about an hour and forty minutes. That is how you make an AI answer from your own documents instead of guessing. Take it if you build.
• Optimize AI Application Performance, 30 minutes on balancing quality, speed and cost. On its own, we would skip it. But you need it for the API certificate, and it is only half an hour.

Track 2: the 3 Codex courses

• Get Started with Codex, 80 minutes. You give Codex a real task in a code repository, check its changes, and show the proof a reviewer needs. If you code, take it.
• Extend Codex Workflows, 90 minutes. You turn your own good habits into shared instructions and reusable skills for your team. Take it if you code with other people.
• Scale Codex Across Governed Teams and Systems, 100 minutes, and the name tells you who it is for: running many Codex jobs in parallel across a company. Only about 1,000 people signed up. If you are not a tech lead, skip it, even though it means no Codex certificate.

Tracks 3 and 4: Leadership, Educators, Students

AI Leadership is the only course in the Lead AI adoption track. Three hours, the longest course in the academy. You pick one AI project at your company and write a real strategy for it: who owns it, the rules and a roadmap. If you manage people, take it. Otherwise, skip it. And remember, it is a badge, not a certificate.

The Teach and learn with AI track has two courses, 45 minutes each. AI for Educators has you build a lesson plan and an assignment from your own materials. AI for College Students covers study plans, group projects and getting ready for job interviews. Take the one that matches your life. Don't take both.

The real number of hours (vs Anthropic)

All fourteen courses add up to about eighteen and a half hours. For comparison, Anthropic's free degree is 28 courses and over 50 hours, with a badge for each course. OpenAI's is smaller, and it adds a certificate when you finish a track. Neither one is an official certification.

Now take out what isn't for you, and the number gets small. If you don't code, it is about four hours for a certificate. If you build, it is about fourteen hours for all three.

The 8-week plan: 3 certificates by Dec 31

1. Week 1 (starts Monday, Nov 9): AI Foundations.
2. Week 2: Applied AI Foundations.
3. Week 3: Agents and Workflows. By Sunday, Nov 29, you have your Foundations certificate. If you don't code, you are done, with a month to spare.
4. Weeks 4 to 6: the three Codex courses, one a week. That is your Codex certificate by Sunday, Dec 20.
5. Week 7: Scope, Evaluate and Optimize.
6. Week 8: Agentic Systems and the retrieval course. On Thursday, Dec 31, you finish the API track.

Three certificates before 2027, at about twenty minutes a day.

Got the certificate. Now what?

Under our last video, someone finished the plan and asked exactly that. Add the certificate to LinkedIn, and call it what it is: a certificate of completion from OpenAI Academy. Then put the thing you built right next to it. The workflow, the test, the agent. A hiring manager can't test a badge. They can look at your work.

Get the free plan

We put the whole plan on one page we call The Honors Plan: every course in order, the minutes, what you need before you start, which ones lead to a certificate, and a box to tick each week. Comment HONORS on the video, or grab it below. Fifty-three days, twenty minutes a day. Pick your track, start on Monday, and finish before 2027.