Every figure on this page is transcribed from NVIDIA’s own certification pages and was last checked on . Prices and policies are set by NVIDIA and can change — verify on nvidia.com before booking. passgenai is an independent study resource and is not affiliated with, endorsed by, or sponsored by NVIDIA Corporation.
Two different objects with similar names. Almost every wrong expectation about NVIDIA training traces back to this row-by-row comparison never being shown.
| DLI course certificate | NVIDIA certification | |
|---|---|---|
| What it proves | You completed a course | You passed an exam |
| Cost | Free on select courses, ~$90 for most paid self-paced ones | $125 to $500 per attempt |
| Supervision | None — you work through it at your own pace | Remote proctoring, webcam and government photo ID |
| Expiry | Does not expire | Two years, then you retake the exam |
| Has an exam code | No | Yes — NCA-GENL, NCP-GENL, NCA-AIIO and so on |
| What employers mean by "NVIDIA-certified" | Not this | This |
Neither is fake and neither is a lesser version of the other — they answer different questions. A course certificate is evidence you studied something; a certification is evidence you can answer questions about it under supervision. If a job posting, partner programme or procurement document asks for a certification, a course certificate will not satisfy it, and that is the only practical consequence of the distinction.
NVIDIA's free catalogue rotates, so this is what each one is for rather than a snapshot of the current list. Check the course page for whether it currently issues a certificate — NVIDIA states that on the page itself.
A short conceptual course on what generative AI is, how large language models work at a level you can explain to someone else, and where the technology is applied and limited. No coding.
Exam relevance: The foundational vocabulary NCA-GENL assumes you already have before its harder domains start. A reasonable first hour if the term "transformer" is not yet concrete for you.
A broader introduction that walks from what machine learning is through to practising with generative models, aimed at people without an AI background — including students and educators.
Exam relevance: Pre-NCA-GENL groundwork. It is wider and shallower than the exam syllabus, so treat it as orientation rather than preparation.
A hands-on course on retrieval-augmented generation and agent composition — embedding, retrieval, orchestration and the LangChain-style plumbing around a model.
Exam relevance: Directly relevant to the RAG domain in NCA-GENL and to a large fraction of NCP-AAI, which is the most agent-specific exam in the programme.
The classic free entry point to GPU programming: kernels, threads, memory, and why moving work to a GPU changes the shape of a problem.
Exam relevance: The GPU-acceleration intuition behind NCP-GENL and the whole AI Infrastructure track. You do not need to write CUDA for any exam, but knowing what the hardware is doing makes several question types obvious.
NVIDIA rotates a set of free applied courses — cybersecurity, healthcare imaging, robotics on Jetson — that are short, self-contained and use real tooling.
Exam relevance: Not exam preparation. Useful for having something concrete to talk about, which is the thing a certification alone cannot give you.
A caution worth stating: no free DLI course is an exam preparation course. They teach a topic; the exams test a published domain list with published weights, and the overlap is partial. Study against the blueprint, and use courses to fill the domains you are weakest in.
Three names for adjacent things, which is why 'NVIDIA Academy login' is a search people run.
The Deep Learning Institute (DLI) is NVIDIA’s training arm and the thing most people are looking for. It runs the self-paced courses — free and paid — and the instructor-led workshops, and it is where your course enrolments and completion certificates live. Course access is through NVIDIA’s learning platform, and you sign in with the NVIDIA account you created when you enrolled.
NVIDIA Academy is a separate, narrower programme: professional and enterprise-oriented training, largely aimed at partners, resellers and enterprise IT staff working with NVIDIA platforms. If you searched for it hoping to find the free generative-AI courses, you want DLI instead.
Certification is a third thing again, administered through NVIDIA’s certification programme rather than either training organisation. Passing an exam is what puts a verifiable badge in your credential wallet; completing a course is what puts a certificate in your DLI profile. They are not the same login flow and not the same record.
The single practical takeaway: if you cannot find a certificate you earned, look in the DLI platform under the account you enrolled with. If you cannot find a certification you passed, look in the credential wallet NVIDIA emailed you after the exam.
Stated plainly, because a page about free things should be clear about its own paywall.
Every exam guide, study plan, prep course lesson, cheatsheet and glossary on this site is free to read, without an account. That covers the published domain weights, what each domain actually tests, a week-by-week plan, and the concepts written out — which is the majority of what you need.
What is paid is the timed mock exams and the advanced practice modes: full-length papers under exam conditions, with explanations for every option. That is the part that costs money to build honestly and the part that is worth paying for, and we would rather say so here than have you discover it three clicks in.
Start with the free exam guides for the four generative-AI exams →
Yes — NVIDIA offers a set of free self-paced courses through its Deep Learning Institute, and select courses issue a DLI certificate of competency on completion. But the important thing to understand is that this is not the same as an NVIDIA certification, and the two are constantly confused. A DLI course certificate is free or bundled with the course, awarded for finishing the material, and unproctored. An NVIDIA certification — NCA-GENL, NCP-GENL, NCA-AIIO and the rest — is a proctored exam costing $125 to $500, sat under webcam supervision, and it is the credential employers and partner programmes mean when they ask whether you are NVIDIA-certified. There is no free route to a certification: no exam in the programme has a zero-cost path other than a voucher. So the accurate answer is that NVIDIA's free courses are genuinely worth taking and will earn you a certificate, but if you need the certification, budget for the exam.
A Deep Learning Institute certificate says you completed a course. An NVIDIA certification says you passed a proctored exam. The first is issued by finishing self-paced material, is available free on select courses, and has no expiry or invigilation. The second costs $125 to $500, is taken online under remote proctoring with photo ID, expires after two years, and carries an exam code such as NCA-GENL that appears on a verifiable badge. Both are legitimate; only the second is what "NVIDIA-certified" means.
No. Every certification exam in the programme has a fee — $125, $200, $400 or $500 depending on the exam — and the only way to sit one at no cost to you is a voucher, typically from a GTC-linked promotion, an employer's Deep Learning Institute allocation, or an instructor-led workshop that includes an attempt. Free NVIDIA *courses* with completion certificates do exist and are a good place to start, but they are a different thing from a certification.
Three tiers, and they stack. NVIDIA's own free self-paced DLI courses cover the fundamentals and issue completion certificates on select courses. Its exam pages publish the domain list and weights for each certification, which is the blueprint you should be studying against. And the foundational papers the exams draw on — the transformer paper, the RAG paper, LoRA, PagedAttention — are free to read and are what the harder questions actually test. On this site the exam guides, study plans, full prep courses, cheatsheets and glossaries are free to read without an account; only the timed mock exams and advanced practice modes are paid.