Is NCA-GENM harder than NCA-GENL? +
Most candidates find it broader, and breadth is what makes it hard. Instead of focusing on text, GENM spans diffusion, vision-language and audio models, plus Multimodal Data (15%) and Performance Optimization (10%), a domain GENL doesn't have. The largest domain is Experimentation at 25%, so memorizing model names isn't enough - you need to know how to run, compare and evaluate experiments. LLM-only preparation leaves the image and cross-modal questions exposed.
How many questions are on NCA-GENM and how long is it? +
NVIDIA lists 50 to 60 multiple-choice questions in 60 minutes, taken online in English through NVIDIA's exam platform (a Certiverse account is required). The domains are Experimentation 25%, Core Machine Learning and AI Knowledge 20%, Multimodal Data 15%, Software Development 15%, Data Analysis 10%, Performance Optimization 10% and Trustworthy AI 5%. Expect about a minute per question, with no time to reason through unfamiliar architectures.
What score do I need to pass NCA-GENM? +
NVIDIA does not publish an official passing score for its Associate exams. Rather than chasing a number, aim to score well above 70% on realistic practice sets in every domain. With seven domains, several of them small, a single blind spot - say Performance Optimization or Multimodal Data - can cost enough questions to tip a borderline result.
How much does NCA-GENM cost and are there prerequisites? +
The exam costs $125. NVIDIA lists no formal prerequisites beyond a basic understanding of generative AI. The certification is valid for two years from issuance, and you recertify by retaking the exam. Since the content shifts as NVIDIA updates its tooling, check the current exam guide before you schedule.
What should I study for the Multimodal Data and Performance Optimization domains? +
Multimodal Data covers preparing, aligning and pairing images, audio and text, including modality-specific preprocessing - candidates used to clean text corpora underestimate how specific these questions are. Performance Optimization is about making multimodal models run efficiently on NVIDIA GPUs, such as reducing precision and speeding up inference. Together they're 25% of the exam, so treat them as core material, not leftovers after model theory.
How long should I prepare for NCA-GENM? +
With a working ML background, 4 to 8 weeks is typical. If you've already passed NCA-GENL, don't assume most of it carries over: spend your time on diffusion and vision-language models, cross-modal data handling, experimentation workflows and GPU performance. If you're new to deep learning, start with the Core ML fundamentals first - they're 20% of the exam and underpin everything else.