AI and machine learning courses from Andrew Ng and industry leaders
DeepLearning.AI is an online learning platform for AI and machine learning education.
AI Panel Score
6 AI reviews
Reviewed
AI Editor ApprovedApproved and published by our AI Editor-in-Chief after full panel analysis.DeepLearning.AI is an online learning platform for AI and machine learning education. Built around content from Andrew Ng and collaborating AI organizations, it offers short courses, professional certificates, specializations, and free resources covering machine learning, deep learning, and applied AI, plus The Batch, a free weekly newsletter; over 7 million people learn with DeepLearning.AI. Hands-on coding labs run in the browser, and the curriculum includes LLM fine-tuning and reinforcement learning alongside structured learning paths. Pro costs $25 per month billed annually or $30 month to month, and a free tier covers course videos; the Team plan is $25 per user per month for 10 to 100 users, while Enterprise and Workforce plans are quoted through sales. TopReviewed's six-seat AI review panel scored it 8.3/10, praising partner instructors from OpenAI, Anthropic, and Meta whose curriculum tracks the frontier, while noting certificates are not universally recognized by technical hiring managers. It best fits teams and individual engineers who need applied AI skills fast.
Learners access DeepLearning.AI primarily through a course catalog organized into individual courses and multi-course specializations. The workflow involves enrolling in a course, progressing through video lectures and hands-on assignments, and completing assessments to earn certificates. Courses are hosted in collaboration with partners and range from foundational machine learning concepts to applied topics like natural language processing and agentic AI workflows.
Beyond the course catalog, the platform distributes free downloadable resources including Andrew Ng's career guide for AI practitioners and the book Machine Learning Yearning, which covers how to structure and tune ML projects. The Batch, a weekly newsletter, provides summaries of current AI research, policy developments, and industry news. These resources are available without enrollment in a paid course.
DeepLearning.AI targets a broad audience from beginners seeking foundational AI literacy to working engineers building production AI systems. Course videos are free to watch, while hands-on labs, quizzes, and certificates require a Pro membership. Organizations can buy the self-serve Team plan for 10 to 100 users at $25 per user per month, billed annually, or custom-priced Enterprise and Workforce plans that add SAML SSO, organization analytics, and a co-branded learning hub.
Courses run in the browser on learn.deeplearning.ai, which DeepLearning.AI now treats as its primary platform and where new courses launch first. A DeepLearning.AI app for iPhone and Android streams lessons and downloads courses for offline viewing. Most courses are also sold on Coursera, where each course or specialization is bought separately and DeepLearning.AI content is not part of Coursera Plus.
In-browser programming environments (including Jupyter notebooks) where learners build and train real AI models, agentic systems, and production applications as part of each course.
Dedicated courses teaching learners how to apply fine-tuning and reinforcement learning techniques to shape model behavior, improve reasoning, and make LLMs safer and more reliable.
A personal dashboard that shows all enrolled short courses and tracks individual learner progress within each course, accessible from the top-right corner on desktop.
A dedicated DeepLearning.AI Forum where learners can ask questions, get peer and instructor support, and share ideas across all courses and specializations.
A paid subscription ($25/mo annually or $30/mo monthly) that unlocks full access to 150+ programs, professional certificates, quizzes, assignments, and new skills added weekly.
A video player with adjustable playback speed, selectable video quality for low-bandwidth users, English/Spanish captions, and a Picture-in-Picture (PiP) mode for multitasking.
Multi-course, in-depth programs (10+ hours) such as the Deep Learning Specialization and Machine Learning Specialization that culminate in a shareable certificate upon completion.
Bite-sized, topic-specific AI courses covering areas such as prompt engineering, RAG, agents, fine-tuning, and LLMOps that learners can complete in one to two hours.
Curated sequences of courses organized by skill level and role (e.g., beginner, AI practitioner, product manager) so learners can follow a progressive path from foundational basics to advanced application.
Course content spans 40+ AI tool providers and frameworks—including LangChain, LlamaIndex, Hugging Face, AWS, Snowflake, MongoDB, and PyTorch—teaching platform-agnostic, production-ready skills.
Courses are taught by instructors from leading AI organizations including OpenAI, Anthropic, Google, Meta, LangChain, CrewAI, and Replit, in addition to Andrew Ng and DeepLearning.AI faculty.
A regularly published AI news and insights newsletter from Andrew Ng covering the latest research, industry trends, and events, delivered to subscribers.
For anyone who wants to watch DeepLearning.AI course videos without hands-on labs, quizzes, or certificates.
For individual learners who want full access to all courses, graded assignments, and certificates on DeepLearning.AI's platform.
Same full Pro access as the monthly plan but billed annually ($300/year), offering the best per-month value for committed learners.
For groups of 10 to 100 people at $25 per user per month, billed annually. The Team Admin invites and removes members and manages the subscription.
For organizations with 100+ users whose technical teams need the full catalog. Custom pricing.
For organization-wide AI fluency across 1,000+ users, with no technical background needed. Custom pricing.
Andrew Ng built the category; $300/year makes this a no-brainer for any AI-serious team.
“DeepLearning.AI is the default choice for structured AI upskilling. Seven million learners and partners like OpenAI, Anthropic, and Google don't happen by accident.”
Founded in 2017 by Andrew Ng, this isn't a startup bet. The Coursera distribution, 7 million learners, and partner roster spanning OpenAI to Hugging Face signal a platform that's built to last. No funding drama to track here.
At $300/year for Pro, you're getting 150+ programs, in-browser Jupyter labs, LLM fine-tuning curriculum, and certificates. Fast.ai is free but unstructured. Udacity's nanodegrees cost 10x this. The pricing doesn't trap you either — audit mode is genuinely free.
The tradeoff: the self-serve Team plan tops out at 100 seats at $25 per user per month, billed annually. If you're trying to move 500 engineers, you're into a custom-priced Enterprise deal with SSO and org analytics, and that's a sales conversation, not a checkout.
Fast.ai and Google's ML crash courses are free but shallow; DeepLearning.AI's partner-led depth at $25/mo is the clear value leader.
OpenAI, Anthropic, and Google as named course partners; no board member will question this choice.
One-to-two hour short courses on prompt engineering and LLMOps can pay back in days, not quarters.
Short courses on LangChain, RAG, and agentic AI workflows directly advance teams building production AI systems today.
Founded 2017, 7 million learners, Coursera-backed distribution, and Andrew Ng's personal brand as a durable moat.
Any team that needs applied AI skills fast and won't pay Udacity prices.
You need enterprise LMS integration or a cohort-based learning structure for large engineering orgs.
Andrew Ng's platform is the default L&D bet for AI upskilling at serious depth.
“DeepLearning.AI offers curriculum breadth that's genuinely hard to match — courses covering 40+ frameworks and providers, partner instruction from Anthropic, OpenAI, and Google, and structured paths from beginner to LLMOps practitioner. At $300/year per seat annually, the cost-per-learning-hour is difficult to argue against.”
150+ programs covering everything from foundational ML to agentic workflow design, with in-browser Jupyter environments baked into the core learning loop. That's not a content library — that's a curriculum architecture. The short-course format (1–2 hours) paired with deep specializations gives L&D leaders real scheduling flexibility across mixed-skill cohorts. The partner instructor roster — Anthropic, Meta, LangChain, CrewAI — means the applied content doesn't lag the industry by 18 months the way most edtech does.
The gap shows up at the organizational layer. The self-serve Team plan runs $25 per user per month, billed annually, but tops out at 100 learners and gives admins seat management, not reporting. Organization analytics, user reporting, and SAML SSO sit in the custom-priced Enterprise plan, and I couldn't find any LMS integration or SCORM export on DeepLearning.AI's business pages. For enterprise L&D with an LMS already in place, that's friction.
If you're building an AI upskilling program against fast.ai or Udacity nanodegrees, DeepLearning.AI wins on curriculum depth and credential weight. The constraint is that learner data and completion workflows live on DeepLearning.AI's own platform, which means they sit in someone else's system rather than your LMS.
7 million learners and Andrew Ng's credential gravity make this the default brand in AI education — fast.ai has community depth, Udacity has nanodegree structure, but neither matches DeepLearning.AI's industry-partner instruction roster.
Short courses and specializations match self-directed practitioner learning, but the Team plan's 100-seat cap and lack of visible LMS hooks limit fit for structured enterprise L&D programs.
No public API and no documented LMS integration; SAML SSO on the Enterprise plan is the main enterprise hook, which is fine for individual learners but constraining for L&D ops teams running SCORM or xAPI workflows.
Weekly course additions and coverage of 40+ frameworks suggest the catalog stays current, but learner data and completion records sit on DeepLearning.AI's platform, outside your stack.
LLM fine-tuning, reinforcement learning, and agentic AI curriculum from actual practitioners at OpenAI and Anthropic — ceiling is genuinely high.
L&D teams building AI upskilling programs for technical and semi-technical staff who need production-ready skills fast.
Your org runs a centralized LMS and needs SCORM-compliant completions or learner data portability.
50 Team seats run $15,000 a year at a published rate, with no quote required.
“The Team plan lists $25 per user per month, billed annually, for 10–100 seats — the same $300 a seat as annual Pro. The catch sits in the fine print: mid-cycle seats are prorated, and removed members earn no refund.”
A 50-person engineering pod can buy this on a card. The Team plan is self-serve: 10–100 seats, $25 per user per month, billed annually. 50 × $25 × 12 = $15,000 a year, $45,000 over three. That's annual DeepLearning.AI Pro pricing, with the Team Admin page at no premium.
Mid-cycle seats are prorated to one shared renewal date. Removed members get no refund and no seat swap. Model turnover as sunk cost. The catch: SAML SSO, organization analytics and user reporting sit in Enterprise, which starts at 100 users, custom-priced. At 50 seats, you're below that line.
Measurement stays per learner: verifiable certificates and the My Learning Progress Tracker. The Free plan streams every course video, so a pilot costs $0. Coursera sells each of these courses separately, and Coursera Plus excludes them. One seat covers 150+ courses.
Self-serve Team checkout, CSV invite import and prorated mid-cycle seats keep procurement light; SAML SSO needs Enterprise.
Pro downgrades keep access to cycle end with progress saved 180 days; Team is annual with no refunds or seat swaps for members removed mid-cycle.
Free, Pro ($30 monthly or $300/year) and Team ($25/user/month, billed annually) are all published; only Enterprise and Workforce are custom-priced.
Verifiable certificates and the My Learning Progress Tracker measure individuals; organization analytics and user reporting are Enterprise-only.
50 Team seats × $300 = $15,000/year, $45,000 over 3 years, with labs, certificates and new courses included.
Engineering teams of 10 to 100 who want a fixed per-seat training budget.
Teams under 100 seats who require SAML SSO.
Andrew Ng's 7-million-learner machine is the default pick for structured AI training
“DeepLearning.AI delivers structured, production-relevant AI curriculum at $25/month annually — hard to beat for individual learners or small teams. The short-course format and Jupyter-in-browser labs keep cohorts engaged without heavy LMS overhead.”
The course architecture is the real differentiator. Short courses at 1–2 hours sit alongside full specializations like the Deep Learning Specialization, so you can slot a lunch-and-learn without rebuilding a whole training calendar. In-browser Jupyter notebooks mean no environment setup emails before class. That alone saves 30 minutes of onboarding friction per cohort session.
Day-3 reality: learners who audit for free hit a wall fast — no graded assignments, no certificates. That free tier works for exploration but won't carry a structured upskilling program. The $25/month annual Pro tier is genuinely reasonable for individuals, and the Team plan carries the same $25 per user per month, billed annually, for groups of 10–100, so a training manager can buy seats without a sales call. fast.ai still edges it for self-directed researchers who want academic depth over career framing.
Courses covering 40+ frameworks and providers — LangChain, Hugging Face, AWS, Anthropic — mean you're teaching tools your learners will actually open on Monday. That's the daily relevance test most corporate training libraries fail.
Short courses and modular paths hold up after the novelty fades, but the audit-tier ceiling means free learners stall without converting to Pro.
Free resources like Machine Learning Yearning and structured learning paths by role signal content built by practitioners, not marketing.
Progress tracker and PiP video player reduce daily annoyances, but groups above the Team plan's 100 seats move to custom Enterprise pricing, which adds procurement drag.
LLM fine-tuning, RLHF, and agentic AI curriculum with industry instructors from OpenAI and Anthropic gives advanced learners real depth to grow into.
In-browser Jupyter labs remove environment friction, and specializations run on the same learn.deeplearning.ai platform as the short courses.
Training managers upskilling engineering or product teams on applied AI and LLM workflows at under $30 per seat.
You need SCORM exports into your own LMS, or SSO and completion reporting without moving to a custom-priced Enterprise plan.
Hands-on labs run on its own platform, and the app covers the commute — on iPhone, not iPad.
“DeepLearning.AI runs its labs, quizzes and certificates on its own platform, and its app downloads lessons for offline viewing. Pro at $25/month billed annually is easy to recommend, but the iOS app has no native iPad build.”
Pro is where this product starts. The Free plan plays every course video, but labs, quizzes and certificates sit behind $30/month, or $25/month billed annually. Pay, and the good stuff opens: in-browser Jupyter labs, saved work you can resume, certificates issued by DeepLearning.AI itself. It all runs on learn.deeplearning.ai, where new courses land first and some, like AI Agents, live only there.
Coursera sells most of these courses separately, and Coursera Plus doesn't cover them. One membership across 150+ courses is the saner deal. The My Learning Progress Tracker is a dashboard, not a coach. Bring your own discipline.
Then the app. It streams lessons, downloads them for the commute, and carries The Batch and Data Points. The catch: the App Store lists it as iPhone-only, so no native iPad build, and I couldn't find the coding labs in it. Hands-on work still wants a laptop.
The video player's Picture-in-Picture, playback-speed and quality controls plus English/Spanish captions are thoughtful, and caption preferences sync between the app and the web.
Short courses handle first-hour; specializations handle month three; role-based learning paths make the jump between them discoverable without hand-holding.
The DeepLearning.AI app streams and downloads lessons offline with progress tracking and certificates, but the iOS build is iPhone-only and I couldn't find the coding labs in it.
The Free plan opens every course video and role-based learning paths answer the 'where do I start' question; labs and quizzes wait behind Pro.
Pro saves your work so you can resume anytime, and the business plans promise progress that resumes across any device; I couldn't find a published uptime figure.
Engineers who want hands-on AI labs with certificates attached.
You need a native iPad app for studying away from a laptop.
Solid courses, but the in-house platform is newly primary and its own learner counts don't match.
“DeepLearning.AI now launches courses first on its own platform, with Pro at $25/month billed annually and a mobile app on version 1.0.5. Certificates are issued by DeepLearning.AI with no accreditation I could find, and its own pages disagree on 7 million versus 10M+ learners.”
Good curriculum. Newer plumbing than the 2017 founding date suggests. The help center now calls learn.deeplearning.ai the primary platform, where new courses launch first. The iPhone and Android app streams lessons, downloads them offline and carries The Batch. Version 1.0.5. Young.
Certificates come from DeepLearning.AI itself, signed by the educators. I couldn't find accreditation behind them. The counting wobbles too: 10M+ learners on the membership page, over 7 million on the homepage. Small. Still makes me read twice.
Exit terms are fair. DeepLearning.AI Pro runs $30 monthly or $25/month billed annually, and a downgrade keeps access until the next billing cycle, with progress saved 180 days. Most courses also sell on Coursera, one at a time. A fallback. The catch: exclusives like AI Agents live only here, so you're betting on the in-house platform.
One Pro subscription covers the full library, while Coursera sells DeepLearning.AI courses one by one and leaves them out of Coursera Plus.
A downgrade keeps access until the next billing cycle and saves progress for 180 days, and most courses also sell on Coursera, though exclusives like AI Agents do not.
The app shipped four updates in September up to version 1.0.5 and a help center covers billing and certificates, though I couldn't find a platform changelog.
Its pages disagree on reach — 10M+ learners on the membership page, over 7 million on the homepage — and the Pro card anchors to a $50.00/mo figure no plan charges.
Named specifics back the pitch — 150+ courses with in-browser labs and quizzes on Pro — and the Free plan's limits (no labs, quizzes or certificates) are stated plainly.
Working engineers who want hands-on GenAI labs under one subscription.
Learners who need an accredited credential rather than a vendor-issued certificate.
Common questions answered by our AI research team
Over 7 million people are learning on DeepLearning.AI.
Courses are created by Andrew Ng and collaborating AI organizations.
Yes, The Batch is a free weekly AI newsletter with news and insights, including Andrew Ng's letter, and it is part of the Free plan.
Yes, free resources include 'How to Build Your Career in AI,' 'Machine Learning Yearning,' and 'A Complete Guide to Natural Language Processing.'
Courses cover machine learning, deep learning, and applied AI, with a focus on building foundational skills and real-world applications.
Company
DeepLearning.AIFounded
2017Pricing
From $25/moFree Plan
Available
DeepLearning.AI is an education technology company based in Palo Alto that offers online courses, specializations, and professional certificates in artificial intelligence and machine learning.
