The companies that build the models now teach them for free. The catch is choosing: there are hundreds of courses, many are outdated, and "free" sometimes means free videos with a paid certificate. Every course below was checked on its official page on October 13, 2026. Pick one path in Step 6, start the first course today, and skip the rest.
1Know what "free" means here
| Label | What it means | Examples |
|---|---|---|
| Free + badge | Course and credential cost nothing | Claude Academy, OpenAI Academy, Google Skills, Microsoft Learn |
| Free + certificate | A downloadable certificate at no cost | Hugging Face Agents Course, Elements of AI |
| Free videos, paid extras | Watch free, pay for labs or the certificate | DeepLearning.AI, CS50 on edX |
| Paid | Free trial, then a subscription or exam fee | Google AI Essentials, Microsoft exams |
On Google Skills, Google's learning hub (Cloud Skills Boost pages now carry its name), the intro AI courses are no-cost, while hands-on labs use credits (developers get 35 free credits a month).
A badge or completion certificate proves you finished a course. It isn't a professional certification. OpenAI says so plainly about its own badges. What employers and clients notice is what you built, so every path below ends with a project.
2Level 1, understand AI and use a chatbot well
Start here if you're new. Pick two, not all of them.
| Course | From | Time | You get |
|---|---|---|---|
| AI Foundations | OpenAI Academy | 60 to 75 min | Badge if you pass the assessment (80%+) |
| Claude 101 | Claude Academy (Anthropic) | 2.5 hr | Completion badge. Works on the Free plan |
| AI Fluency: Framework and foundations | Claude Academy | 4 hr | Completion badge. Not tied to one product |
| Introduction to Generative AI | Google Skills | 45 min | Badge |
| Introduction to AI | Elements of AI (University of Helsinki) | Self-paced, about 6 weeks suggested | Free certificate. No complicated math or code |
| Generative AI for Everyone | DeepLearning.AI | 5 hr | Free videos |
Google AI Essentials is often listed as free, but it isn't: in the US and Canada it costs $49 a month on Coursera after a 7-day free trial. It takes under 5 hours, so you can finish it inside the trial. Set a reminder to cancel if you don't want to keep paying.
3Level 2, use AI at work with the tool you picked
Match the course to the assistant you chose in Choose your AI stack.
| Course | From | Time | You get |
|---|---|---|---|
| Applied AI Foundations | OpenAI Academy | 75 to 90 min | Badge. Turns one recurring task into a workflow |
| Agents and Workflows | OpenAI Academy | 75 to 90 min | Badge. Needs ChatGPT Work access |
| Introduction to Claude Cowork | Claude Academy | 2.5 hr | Completion badge. Hands-on parts need a paid plan |
| AI Fluency for small businesses | Claude Academy | 8 lessons and a quiz | No badge listed |
| Google Workspace with Gemini | Google Skills | 11 activities, starting with a 30-minute foundations course | Badge for the foundations course |
| AI Prompting for Everyone | DeepLearning.AI | 7 hr | Free videos |
| Microsoft Copilot training | Microsoft Learn | Modules | Achievement badges when signed in |
4Level 3, build with APIs, MCP and coding agents
You'll need a computer and, for the API courses, a few dollars of API credit.
| Course | From | Time | You get |
|---|---|---|---|
| Claude Code 101 and Claude Code in action | Claude Academy | 1.5 hr + 1 hr | Badges. Claude Code needs Pro or higher |
| Introduction to MCP and MCP: Advanced topics | Claude Academy | 1 hr + 1.5 hr | Badges |
| Building with the Claude API | Claude Academy | 9 hr | Badge |
| Get Started with Codex, Design and Build Agentic Systems, Build with RAG (list) | OpenAI Academy | 80, 110 and 100 min | Badges |
| MCP with Anthropic, Claude Code, Prompt Engineering for Developers | DeepLearning.AI | 1 hr 40 min to 2 hr each | Free videos |
| Generative AI for Beginners | Microsoft (GitHub) | 21 lessons | Free, open source, no certificate |
| 5-Day AI Agents Intensive | Google and Kaggle | 5 days of papers and codelabs | No-cost. Needs a phone-verified Kaggle account and an AI Studio key |
| Grok API quickstart | xAI docs | Short | No course or badge. Docs only |
5Level 4, go deeper on agents and models
| Course | From | Time | You get | Needs |
|---|---|---|---|---|
| Agents Course | Hugging Face | 3 to 4 hr a week per unit | Two free certificates: fundamentals after Unit 1, completion after an assignment and the final challenge | Basic Python |
| LLM Course | Hugging Face | 6 to 8 hr a week per chapter | No certificate | Good Python |
| Agentic AI | DeepLearning.AI | About 10 hr | Free videos. Labs and certificates need Pro, $300 billed yearly | Python |
| AI Agents for Beginners | Microsoft (GitHub) | 18 lessons | Free, no certificate | Python |
| Practical Deep Learning for Coders | fast.ai | 9 lessons of about 90 min | Free | About a year of coding |
| CS50's Introduction to AI with Python | Harvard | 7 weeks | Free. edX sells a verified certificate | CS50x or a year of Python |
Microsoft Learn training is free and needs no account, but the Azure AI Fundamentals exam (AI-901) is paid, priced by country. Start with its free Introduction to AI concepts module.
6Pick one path by goal
| You are a... | Your path, in order | Time | Finish with |
|---|---|---|---|
| Beginner | AI Foundations, then AI Fluency, then Google's Introduction to Generative AI | About 6 hr | One weekly task you now do with AI |
| Founder | Claude 101, Applied AI Foundations, Claude Code 101, Introduction to MCP | About 6.5 hr | A workflow for your busiest weekly task, and a prototype |
| Business owner | AI Foundations, then the Gemini foundations course (if you run on Google Workspace) or Claude 101, then Agents and Workflows | 3 to 5 hr | One process handed to an agent, with your review step |
| Creator or marketer | AI Prompting for Everyone, Applied AI Foundations, then AI Fluency for creative work | About 8.5 hr, plus 8 short lessons | A repeatable content workflow and your own prompt library |
| Developer | Claude Code in action, Introduction to MCP, Building with the Claude API (or OpenAI's Codex and agent courses), then the Hugging Face Agents Course | About 11 hr, then 3 to 4 hr a week | A working agent in a public repo, with no keys in it |
7Make it stick
Courses only pay off when you use them on real work the same week. Book one slot a week, finish one course at a time, and apply each lesson to a task from your own job.
I want to learn AI for this goal: [goal, e.g. "automate my client reporting"]. My level: [beginner / use ChatGPT daily / can code in Python]. Time: [hours] per week. From this list only, pick 3 to 5 courses in order and explain why each one: [paste the rows from the guide that fit you] For each course, give me one task from my real work to practice on that week. Don't add courses that aren't on my list.
What's next
Put the Claude courses to work with Set up Claude and Claude Projects. Developers can go straight to Ship a landing page with Claude Code and Connect Claude to anything with MCP.