The Best AI Certifications in 2026, Ranked: Which Should You Get?

The best AI certification for most people in 2026 is the AWS Certified AI Practitioner ($100), or Microsoft Azure AI Fundamentals (AI-901, $99) if your employer runs on Microsoft. After that first step, pick by goal: Google Cloud Generative AI Leader for managers, Microsoft AI-103 or the AWS Generative AI Developer Professional for developers, the AWS Machine Learning Engineer Associate or Google’s Professional Machine Learning Engineer for ML engineers, CompTIA SecAI+ for security, IAPP AIGP for governance, and Oracle OCI AI Foundations if you want one that costs nothing.

We compared 34 AI certifications from AWS, Microsoft, GitHub, Google Cloud, NVIDIA, Oracle, CompTIA, CertNexus, Databricks, Salesforce, PMI, IAPP, ISACA, VMware by Broadcom and Anthropic, using only each vendor’s own exam pages. Start with the top 10 ranked, scan the full comparison table, or jump to the best pick for your goal. Several exams retired, changed or launched in 2026, so check the retired and changing list before you study from an older guide.

The top 10 AI certifications for 2026, ranked

This ranking is for the typical reader: certifications anyone can take, from platforms employers actually run, current for 2026, and priced fairly for what they prove. Your role can reorder it, so check the best pick for your goal before you book.

1. AWS Certified AI Practitioner (AIF-C01), $100. The best first AI certification: no coding, no prerequisites, valid for 3 years, and updated in April 2026 for agentic AI.
2. Microsoft Azure AI Fundamentals (AI-901), $99. The best first certification in a Microsoft shop, and it never expires. Expect basic Python.
3. Google Cloud Generative AI Leader, $99. The best pick for managers and business leaders, with no technical background required.
4. Microsoft Azure AI Apps and Agents Developer (AI-103), $165. The best developer certification for building generative AI apps and agents.
5. AWS Machine Learning Engineer Associate (MLA-C02), $75 in beta. The best for ML engineers; the new version adds generative AI, RAG and agents. AWS associate exams are normally $150.
6. NVIDIA Generative AI LLMs (NCA-GENL), $125. The best cloud-neutral technical certification: how large language models work, not one vendor’s console.
7. AWS Generative AI Developer Professional (AIP-C01), $300. The best advanced certification for senior generative AI developers.
8. CompTIA SecAI+ (CY0-001), $298. The best for security professionals who protect AI systems. It isn’t entry level.
9. IAPP AIGP, $649 for members, $799 otherwise. The best for AI governance, privacy, risk and compliance roles.
10. Oracle OCI AI Foundations Associate, free. The best free AI certification, exam included, and most useful if your company runs Oracle.

Just missed the cut: GitHub Copilot (GH-300) for developers who code with an AI assistant, Google’s Professional Machine Learning Engineer for senior ML engineers on Google Cloud, PMI-CPMAI for project managers, and Anthropic’s Claude certifications, which would rank higher if they weren’t limited to Claude partner companies. Seven of the ten above have free practice questions in our AI exam prep hub, so you can test the level before you pay.

The 34 AI certifications we rank, compared

Checked against vendor pages on October 8, 2026. Prices are US list prices before tax, and most vendors charge different amounts in other countries. “Not stated” means the vendor’s exam page doesn’t say, so we don’t guess. For the full bill, including retakes and renewals, see what an AI certification actually costs or run the numbers in the AI certification cost calculator.

Every AI exam, side by side (live table)

The table above covers the 34 certifications we rank: the ones most people should consider. For every AI exam we track, including specialist, security and retired exams, with exam length, passing score and the date each row was last checked against the vendor, use the live table below. It updates itself whenever we re-check an exam.

Want filters by vendor and the option to hide retired exams? Open the full AI Certification Requirements Index.

The best AI certification for each goal

Short on time? Here’s one pick for each job. The details for every role follow below.

Best for beginners: AWS Certified AI Practitioner

$100, no prerequisites, no coding, valid for 3 years, and its exam guide was updated in April 2026 to cover agentic AI and MCP. It’s open to anyone and it sits at the bottom of a clear path: passing the ML Engineer Associate or the Generative AI Developer Professional later renews it automatically. The other no-prerequisite starters are AI-901 ($99), Google’s Generative AI Leader ($99) and Oracle’s free OCI AI Foundations. If you’re starting from zero, our guide to AI certifications for beginners lays out the order to take them in, and ten minutes of free AIF-C01 practice questions will show you the level.

Best if you work in a Microsoft shop: Azure AI Fundamentals (AI-901)

$99 in the US, and Microsoft Fundamentals certifications never expire. It replaced AI-900 in 2026 and is more hands-on than the old exam: more than half of it is building with Microsoft Foundry, and Microsoft expects basic Python. Torn between this and AWS? Read AWS AI Practitioner vs Azure AI Fundamentals, and see Microsoft AI certifications in 2026 for the full Microsoft path.

Best free AI certification: Oracle OCI AI Foundations

Oracle includes the OCI AI Foundations Associate in its free Foundations training and certification program, so the course and the exam cost nothing. It’s the only major vendor AI cert on this list with no exam fee. It counts for most if your company runs Oracle. More options in free AI certifications: what’s actually free.

Best for managers and business leaders: Google Cloud Generative AI Leader

$99, no prerequisites, valid for 3 years, and Google aims it at any job role, technical or not. It covers generative AI basics, Google Cloud’s AI offerings and the business strategy behind an AI rollout. In a Microsoft company, the AI Transformation Leader (AB-731, $99) covers similar ground. AWS’s new AI Business Strategist (AIB-C01) is $50 while it’s in beta.

Best for non-technical business users: Microsoft AI Business Professional (AB-730)

$99 and 45 minutes, for people who use generative AI tools such as Microsoft 365 Copilot in their daily work and will never build an AI app. If you’re a student and your school offers it, CompTIA AI Fundamentals is the vendor-neutral alternative.

Best for developers and AI engineers: Microsoft AI-103 (Azure) or AWS AIP-C01 (AWS)

On Azure, AI-103 ($165) is built around generative AI apps and agents, its biggest domain at 30–35%. AI-200 ($165) suits developers who build the cloud side of AI apps (containers, vector databases, event-driven pipelines), and AI-500 is the expert step after AI-103 for multi-agent systems. On AWS, the Generative AI Developer Professional (AIP-C01, $300) is the senior option. If your company is a Claude Partner Network member, the Claude Certified Developer: Foundations (CCDV-F) fits developers building on Claude.

Best for coding with an AI assistant: GitHub Copilot (GH-300)

$99, 100 minutes, valid for 2 years, and open to anyone with GitHub basics and some programming experience. It tests Copilot’s features, prompt and context crafting, privacy settings and responsible use, so it suits developers whose team already pays for Copilot. GitHub’s newer Agentic AI Developer exam (GH-600, $165) goes further into coding agents, MCP servers and multi-agent workflows. Our GH-300 study guide walks through the published skills outline.

Best for building AI agents: Model Context Protocol Associate (MCPA)

MCP (Model Context Protocol) is the open standard that lets AI assistants connect to your tools and data, and it already shows up in exams like the AWS AI Practitioner and GitHub’s GH-600. The Linux Foundation and the Agentic AI Foundation, which hosts the MCP project, launched this vendor-neutral exam on September 14, 2026. It costs $250 for a 90-minute, multiple-choice exam taken online with a proctor, with one retake included. It’s valid for 2 years and has no prerequisites. Our MCP certification guide covers its five domains.

Best for ML engineers and MLOps: AWS Machine Learning Engineer Associate

$150 standard, valid 3 years, and the new MLA-C02 version adds generative AI, RAG and agentic AI. If you work on Google Cloud, take the Professional Machine Learning Engineer ($200) instead; Google recommends 3+ years of industry experience. If you run models in production on Azure, it’s AI-300 ($165). Experienced data scientists who want a vendor-neutral credential can look at CompTIA DataAI ($544, formerly DataX), an expert-level exam for people with about five years in data science.

Best cloud-neutral technical cert: NVIDIA NCA-GENL

$125 for a one-hour exam on how large language models work, not on one cloud’s console. Infrastructure people should look at its sibling, NCA-AIIO, which is 40% AI infrastructure. Our NCA-AIIO vs NCA-GENL comparison helps you choose.

Best for security professionals: CompTIA SecAI+

$298, launched February 17, 2026, for people who secure AI systems. It isn’t entry level: CompTIA recommends 3–4 years in IT and at least 2 years hands-on in cybersecurity. If you already hold CISM or CISSP, ISACA’s AAISM is the management-level option. In a Microsoft shop, the new Cloud and AI Security Engineer Associate (SC-500, $165) replaces AZ-500 and covers securing Azure workloads, including AI. Our AI security certifications guide compares all of them.

Best for AI governance, privacy and risk: IAPP AIGP

$649 for IAPP members or $799 otherwise, for legal, privacy, risk and compliance people who govern AI rather than build it. IT auditors who hold CISA (or a qualifying audit credential) should look at ISACA’s AAIA, and risk professionals with a qualifying certification can take ISACA’s AAIR (Advanced in AI Risk), launched in April 2026. See the AIGP study guide and our AI governance certifications comparison.

Best for project managers: PMI-CPMAI

From the organization behind the PMP. It tests a method for running AI and data projects, not coding, and PMI requires no prior project management or AI experience. You take PMI’s prep course before you can book the exam. Our PMI-CPMAI study guide covers the published outline.

Best for Salesforce or Databricks shops: the platform’s own cert

Salesforce Agentforce Specialist ($200) for building agents on Salesforce, or Databricks Generative AI Engineer Associate ($200) if your data lives in Databricks. Both are strong inside their platform and carry little weight outside it.

Best for AI consultants and solution architects: the expert cert for your clients’ platform

There’s no single AI consultant certification. Clients hire for the platform you’ll build on, so take the senior exam for that platform: Microsoft’s Agentic AI Business Solutions Architect Expert (AB-100, $165) for agent solutions on Microsoft’s business apps, AWS AIP-C01 for AWS, or the Claude Certified Architect exams if your firm is a Claude partner. If you advise on strategy rather than build, pair it with a business credential such as Google’s Generative AI Leader or AWS’s AI Business Strategist.

Best at a Claude partner firm: Claude Certified Architect: Foundations (CCAR-F)

$125 and the original Claude exam. Take the Associate (CCAO-F) if you don’t write code, or the Developer (CCDV-F) if you build against the API. These exams are only open to Claude Partner Network employees; see who can take the Claude certification before you study, and the Claude certification guide for all four exams.

Prefer video? The 8-minute ranking

This video ranks the 11 major AI certifications as they stood in July 2026. The written guide on this page is kept current and now covers 34, so where the two differ (for example, MLA-C01’s English retirement and AWS’s new business and developer certs), go by the page.

How to choose: three questions

A certification only pays off if it passes three tests for your situation:

1. Recognition. Will the person you want to hire you, or promote you, know this certification? A credential nobody recognizes is a certificate of attendance.
2. Stack match. Most companies run on one cloud or one platform. A cert in the platform your employer actually uses is worth far more than one in a platform it doesn’t.
3. Role match. Do you lead people who use AI, or build with it yourself? Leadership certs and engineering certs test different things.

If a cert fails two of the three for you, skip it, however good it is for someone else. Choosing for a whole team instead? Which AI certification your IT team should get first has a role-by-role matrix.

Every AI certification, in detail

Entry level: no experience required

• AWS Certified AI Practitioner (AIF-C01): $100. 65 questions (50 scored) in 90 minutes, pass at 700 on a 100–1,000 scale, valid for 3 years. AWS’s exam guide version 1.1 (April 30, 2026) added agentic AI, MCP and context engineering, and the largest domain is applications of foundation models (28%). No coding. Study guide · Free practice questions · Exam cost · Is it worth it? · 3-week study plan · How long to study
• Microsoft Azure AI Fundamentals (AI-901): $99. Passing score 700, and Microsoft’s Fundamentals exams run 45 minutes. It doesn’t expire. AI-901 replaced AI-900 on June 30, 2026: 55–60% of the exam is implementing AI solutions with Microsoft Foundry, and Microsoft expects basic Python. Study guide · Free practice questions · Exam cost · Is it worth it? · 2-week study plan · AI-900 vs AI-901
• Google Cloud Generative AI Leader: $99. 50–60 multiple-choice questions in 90 minutes, no prerequisites, valid for 3 years, offered in English, Japanese, Spanish and Portuguese. It covers generative AI fundamentals, Google Cloud’s generative AI offerings, techniques to improve model output, and business strategy for a generative AI rollout. Google recently updated the exam for product renames, so check names against the current exam guide. Study guide · Free practice questions
• Oracle OCI AI Foundations Associate (1Z0-1122-26): free. Part of Oracle’s free Foundations training and certification program, delivered through Oracle MyLearn. It introduces AI, machine learning and generative AI and how they’re used on Oracle Cloud Infrastructure. The 2026 exam is 60 minutes online, proctored by AI through a browser extension and your webcam, and the certification is valid for 24 months. Oracle hadn’t published the question count or passing score for this version on the pages we checked. Study guide · Free practice questions · What’s actually free
• CompTIA AI Fundamentals: $189 course, or $99 per learner through schools. A three-credit course (50–56 instructional hours) with a built-in assessment that earns a CompTIA AI Fundamentals CompCert, a competency certificate rather than a proctored certification exam. It covers prompt design, responsible AI use, checking AI output and automation basics, with no technical background required. CompTIA says plainly that it is not a certification. Best if your school or college offers it. Study guide · Free practice questions
• Microsoft AI Business Professional (AB-730): $99. Beginner level, 45 minutes, proctored. For business users who use generative AI tools such as Microsoft 365 Copilot and its agents in daily work, without writing code or building apps. Microsoft updates the English exam on October 20, 2026, so check the change list in its study guide.
• Microsoft AI Transformation Leader (AB-731): $99. Beginner level, 45 minutes, no coding. For business decision-makers who lead AI adoption with Microsoft 365 Copilot and Microsoft Foundry; Microsoft expects experience leading adoption or change management. It’s the Microsoft-shop counterpart to Google’s Generative AI Leader.
• AWS Certified AI Business Strategist (AIB-C01): $50 in beta, $100 standard. Announced September 1, 2026, with beta exams from September 29. The beta is 85 questions in 170 minutes (the standard exam is 130 minutes), pass at 700, valid for 3 years, in English and Japanese for now. No coding or hands-on AWS work; AWS suggests about six months working with or alongside teams adopting AI. The largest domain is AI strategy and business value (28%). Earn it by February 15, 2027 and AWS adds an Early Adopter badge. AIB-C01 guide

Intermediate: build, deploy and run AI

• NVIDIA Generative AI LLMs Associate (NCA-GENL): $125. 50–60 multiple-choice questions in one hour, taken online through Certiverse, valid for 2 years (recertify by retaking it). Topics: core machine learning and AI knowledge 30%, software development 24%, experimentation 22%, data analysis 14%, trustworthy AI 10%. Not tied to any one cloud. Study guide · Free practice questions · Exam cost · Is it worth it?
• NVIDIA AI Infrastructure and Operations Associate (NCA-AIIO): $125. 50 questions in one hour, valid for 2 years. AI infrastructure 40%, essential AI knowledge 38%, AI operations 22%. NVIDIA recommends a basic understanding of data center infrastructure. Study guide · Free practice questions · Exam cost · NCA-AIIO vs NCA-GENL
• Microsoft Azure AI Apps and Agents Developer Associate (AI-103): $165. 120 minutes, pass at 700, and like all Microsoft associate certs it expires after a year but renews free with an online assessment. Implementing generative AI and agentic solutions is the largest domain (30–35%), and Microsoft expects experience building apps in Python. It replaced AI-102, which retired on June 30, 2026. Study guide · Free practice questions · Exam cost · Is it worth it? · 3-week study plan · AI-102 vs AI-103 · AI-901 vs AI-103
• Microsoft Azure AI Cloud Developer Associate (AI-200): $165. Intermediate, 120 minutes. It covers building and running AI solutions on Azure with containers, vector-enabled databases, event-driven AI pipelines, serverless functions, secret management and distributed observability. Microsoft announced the beta in May 2026, its June roundup put general availability in July 2026, and it lists AI-200 as the replacement for the AZ-204 Azure Developer Associate.
• Microsoft Machine Learning Operations Engineer Associate (AI-300): $165. 120 minutes, English only, pass at 700, free yearly renewal. It covers running machine learning and generative AI in production: model lifecycle and operations (25–30%), GenAIOps infrastructure (20–25%), plus quality, observability and optimization. Microsoft expects a data science background, Python and entry-level DevOps (GitHub Actions, CLIs, Bicep). It replaced DP-100, retired June 1, 2026. Study guide · Is it worth it? · What replaced DP-100
• Microsoft Cloud and AI Security Engineer Associate (SC-500): $165. 120 minutes, pass at 700, free yearly renewal. It covers identity, access and governance (20–25%), securing storage, databases and networking (25–30%), securing compute, and managing security posture, and Microsoft lists it as the successor to AZ-500 (Azure Security Engineer Associate). Microsoft expects hands-on Azure and hybrid administration and strong Entra ID skills.
• GitHub Copilot (GH-300): $99. 100 minutes, pass at 700, valid for 2 years, through Pearson VUE online or at a test center, with no Microsoft Learn access during the exam. GitHub expects GitHub fundamentals and experience with at least one programming language. Using Copilot features is the largest domain (25–30%), then responsible use (15–20%); skills measured as of August 7, 2026. GitHub is moving its certifications to Microsoft’s renewal process. Study guide
• AWS Certified Machine Learning Engineer Associate (MLA-C02 and MLA-C01): $150. The English MLA-C01 exam ended on September 28, 2026 (65 questions, 130 minutes, pass at 720); Japanese, Korean and Simplified Chinese versions continue until MLA-C02 goes live. MLA-C02 has been in beta since September 29, 2026: English only, $75, 85 questions in 170 minutes. AWS’s September announcement puts general availability, and MLA-C01’s full retirement, on January 14, 2027. MLA-C02 adds generative AI with Amazon Bedrock and RAG, agentic AI, and foundation model selection and fine-tuning. Valid for 3 years, including beta passes, and earning it renews the AI Practitioner too. MLA-C02 study guide · MLA-C02 practice questions · What changed from MLA-C01 · MLA-C01 study guide · MLA-C01 practice questions · Exam cost · Is it worth it? · 6-week study plan
• Databricks Certified Generative AI Engineer Associate: $200. 45 scored questions in 90 minutes, online or at a test center, valid for 2 years (recertify by passing the current version). Databricks recommends 6+ months of hands-on experience; application development is the largest section (30%), and all code on the exam is Python.
• Salesforce Certified Agentforce Specialist: $200 (retake $100). 60 questions plus up to five unscored in 105 minutes, 72% to pass, no prerequisite. Salesforce pictures a candidate with about a year of configuration experience, including Agent Builder and Prompt Builder. AI agents are 35% of the exam. You keep it current with a yearly maintenance module on Trailhead. Study guide
• CertNexus Certified Artificial Intelligence Practitioner (CAIP, AIP-210): $367.50. 80 scored plus 10 trial questions in 120 minutes, pass at 60% or 59% depending on the exam form. CertNexus describes the target candidate as having 1 to 3 years of professional exposure and recommends Python or R and statistics. Operationalizing ML models is the largest domain (30%). Valid for 3 years; renew by retaking the current exam or through CertNexus’s continuing education program (90 credits and a $150 fee per three-year cycle). CertNexus is a separate company from CompTIA. Study guide · Free practice questions · Is it worth it?
• VMware VCP-VCF Administrator (2V0-17.25): $250. 60 questions in a 135-minute appointment, pass at 300 (scaled), valid for 3 years. Broadcom recommends a year in IT plus six months with VMware Cloud Foundation. It isn’t an AI exam; it’s here because private-cloud teams ask about it alongside AI certs. Skip it unless you run VCF. Study guide · Free practice questions

Advanced, professional and specialist

• AWS Certified Generative AI Developer Professional (AIP-C01): $300. 75 questions (65 scored) in 180 minutes, pass at 750, valid for 3 years. The largest domain is foundation model integration, data management and compliance (31%), and AWS refreshed the exam in March 2026 to include Amazon Bedrock AgentCore. Passing it also renews the AI Practitioner and the ML Engineer Associate. Study guide · Exam cost
• Google Cloud Professional Machine Learning Engineer: $200. 50–60 multiple-choice and multiple-select questions in two hours, in English or Japanese, valid for 2 years. Google recommends 3+ years of industry experience, including at least a year designing and managing solutions on Google Cloud. The exam was updated for the move from Vertex AI to Gemini Enterprise Agent Platform, so older study material uses outdated product names.
• Microsoft Multi-Agent AI Solutions Expert (AI-500). An expert-level exam on designing, building and operating production multi-agent systems with Microsoft Foundry and Azure, including MCP and RAG. You also need the AI-103 certification to earn it. Microsoft put the beta in July 2026 and general availability in September 2026; it is English only and costs $165 in the US.
• Microsoft Agentic AI Business Solutions Architect Expert (AB-100): $165. Expert level, generally available since February 2026. To earn the credential you pass AB-100 and hold one of 13 Microsoft associate certifications, such as AI-103, AI-300, AB-620 or PL-400. Deploying AI-powered business solutions is the largest domain (40–45%), with planning and designing them at 25–30% each. Microsoft updates the English exam on October 14, 2026. Free yearly renewal.
• CompTIA SecAI+ (CY0-001): $298, or $332 with a retake. Launched February 17, 2026. Up to 60 multiple-choice and performance-based questions in 60 minutes; you need 600 on a 100–900 scale. English and Japanese. CompTIA recommends 3–4 years in IT and 2+ years of hands-on cybersecurity, with Security+, CySA+ or PenTest+ as a typical background. Renew through continuing education, with a $75 fee per three-year cycle. Study guide
• CompTIA DataAI (DY0-001): $544, or $709 with Retake Assurance. CompTIA renamed DataX to DataAI on January 21, 2026. Up to 90 multiple-choice and performance-based questions in 165 minutes, scored pass/fail, in English and Japanese. CompTIA recommends 5+ years in data science, and machine learning and modeling are the largest domains (24% each). Renew with 75 CEUs and a $150 fee every three years; CompTIA estimates the exam will retire in 2027. CompTIA’s AI certifications compared
• PMI Certified Professional in Managing AI (PMI-CPMAI): price varies. 120 questions (100 scored) in 160 minutes, online or at a Pearson VUE center. You must finish PMI’s CPMAI exam prep course before you can schedule, but no prior project management or AI experience is required. PMI sets the course and exam fee by region and membership. The two biggest domains, identifying business needs and identifying data needs, are 26% each. Keep it with 30 PDUs every three years. Study guide
• IAPP Artificial Intelligence Governance Professional (AIGP): $649 members, $799 non-members. 100 questions (85 scored) in 2.75 hours. The certification term is two years; to renew you submit 20 continuing education credits and pay a maintenance fee (included for IAPP members, $250 a term for non-members). For privacy, legal, risk and compliance roles. Study guide
• ISACA Advanced in AI Security Management (AAISM): $459 members, $599 non-members. Only open to people who already hold CISM or CISSP. 90 multiple-choice questions in 2.5 hours, pass at 450 on a 200–800 scale. Domains: AI governance and program management 31%, AI risk and opportunity management 31%, AI technologies and controls 38%. Keep it with 10 CPE hours a year (30 over three years) and a small yearly fee. Study guide
• ISACA Advanced in AI Audit (AAIA): $459 members, $599 non-members. For IT auditors who hold CISA or a qualifying audit credential such as CIA or CPA. 90 multiple-choice questions in 2.5 hours. AI operations is the largest domain (46%). It requires 10 hours of AI-focused CPE each year.

Partner-only: Anthropic’s Claude certifications

Anthropic launched its first Claude certification (CCAR-F) on March 12, 2026 and announced three more on July 23, 2026. All four share the same rules: 120 minutes, pass at 720 on a 100–1,000 scale, English only, delivered by Pearson VUE online or at a test center, and valid for 12 months, with a free renewal if you renew on time.

• Claude Certified Associate: Foundations (CCAO-F): $99. 60 questions. For people who use Claude as a work tool in operations, marketing, project management, education and communications, not for developers. It’s the only one of the four that doesn’t count toward a partner’s network tier. Study guide · Free practice questions · 3-week study plan
• Claude Certified Developer: Foundations (CCDV-F): $125. 53 questions. For engineers who build, integrate and ship Claude apps and agents; applications and integration is the largest domain (33.1%). Study guide · Free practice questions
• Claude Certified Architect: Foundations (CCAR-F): $125. 60 questions built on four scenarios drawn from a bank of six. Agentic architecture and orchestration is the largest domain (27%). The price rose from $99 on June 30, 2026. Study guide · Free practice questions · 4-week study plan
• Claude Certified Architect: Professional (CCAR-P): $175. 63 questions, for mid- to senior-level architects delivering Claude at enterprise scale. Anthropic recommends 3+ years in systems architecture, and you don’t need CCAR-F first. Study guide · Free practice questions · 6-week study plan · CCAR-F vs CCAR-P

The catch: only employees of Claude Partner Network companies can register, using a work email on a recognized company domain. Joining the network is free, but new partners start at the Registered tier with a commitment of at least 10 certified practitioners, which rules out most freelancers and small shops. Higher partner tiers get 50% off, and Global Premier partners get 100% off through December 31, 2026. Read who can actually take the Claude certification, what the Claude exams cost and whether it’s worth it before you study.

Retired and changing in 2026: check before you study

A lot of AI certification advice online is out of date. These are the changes that matter right now:

Already hold a retired cert? Retirement doesn’t revoke it: it stays valid until its own expiry date. For every cert’s renewal rules, see do AI certifications expire? and the free AI certification renewal tracker.

When a certification is a waste of money

1. Stacking entry-level certs. AWS AI Practitioner, Azure AI Fundamentals and Google’s Generative AI Leader cover a lot of the same ground under different logos. The second one adds little. Pick one, then move up a level or go build something.
2. Passing without practice. If you pass but never actually use the tools, the first technical interview will show it. A cert proves you learned something; it doesn’t replace doing it.
3. The wrong-stack cert. An Oracle cert in an AWS shop, or a Salesforce cert when you’re aiming at a data engineering role. No exam score fixes a platform mismatch.

Wondering whether any of this moves your paycheck? Read do AI certifications increase your salary?

How we ranked these certifications

The top 10 ranks across all 34 for a typical reader who can sit any open exam. The best-by-goal picks rank within each goal, because “best” depends on what you need the certification to do. For every cert, we checked six things on the vendor’s own pages:

1. Who can take it. Open to anyone, gated behind another credential (ISACA’s AAISM and AAIA), or limited to partner employees (Anthropic). Open certs rank higher for general readers.
2. Total cost. The US list price, plus what it costs to stay certified: free renewal, a retake, or continuing education with a fee.
3. Validity. From 12 months (Claude, Microsoft associate certs) to never expiring (Microsoft Fundamentals).
4. Level and prerequisites. The vendor’s own label and recommended experience. The Level column is our summary of those. A cert you can pass and use now ranks above a harder one you can’t use yet.
5. Currency. Whether the exam was updated in 2026 for agents, RAG and current product names, or is being retired or replaced.
6. Platform and role fit. Whether it’s tied to one cloud or platform. Platform certs rank highest for people on that platform and drop for everyone else.

What we didn’t use: pass rates (none of the vendor pages we checked publish them), salary figures, or difficulty scores we can’t measure. Where a vendor page doesn’t state a fact, the table says so. HOW TO // AI sells practice questions for some of these exams; to keep that from tilting the picks, every cert is judged on the same six criteria, including the many we don’t cover, such as AWS AIP-C01, Google’s Professional ML Engineer, GitHub GH-300, CompTIA SecAI+ and IAPP AIGP.

Checked against vendor pages on October 8, 2026: AWS Certification and AWS exam guides, Microsoft Learn (including GitHub’s certification pages) and the Microsoft Skills Hub blog, Google Cloud certification pages and help center, NVIDIA certification pages, Oracle University, CompTIA, CertNexus, Databricks, Salesforce, PMI, IAPP, ISACA, Broadcom, and Anthropic’s Partner Academy and exam guides.

Your next step

Down to two? The fastest tiebreaker is ten minutes of practice questions on each: you’ll feel which material clicks. Every exam with a “Free practice questions” link above has a free, no-sign-up set in our AI certification exam prep hub, and the free study plan generator turns your exam date into a week-by-week plan. On iPhone, the same questions are in our free AI certification practice app.

Want a five-minute gut check first? Open the official exam guide, copy the task list for its biggest domain, and paste it into ChatGPT or Claude with this prompt: “Quiz me with five questions on this, one at a time, and explain each answer.” Five minutes later you’ll know whether you’re close or just getting started. The same prompt works for anything you need to learn, from a training doc to a new policy.

HOW TO // AI is not affiliated with or endorsed by Amazon Web Services, Microsoft, GitHub, Google, NVIDIA, Oracle, CompTIA, CertNexus, Databricks, Salesforce, PMI, IAPP, ISACA, Broadcom/VMware or Anthropic. All certification names are trademarks of their respective owners, referenced descriptively. Prices are US list prices checked on each vendor’s site on October 8, 2026; confirm on the official exam page before you book.

Keep exploring: AI Certification Requirements Index · AWS AI certifications · Microsoft AI certifications · CompTIA AI certifications · free AI certifications · generative AI certifications · AI governance certifications · AI security certifications · Claude Certified Architect · AIGP · SecAI+ · GH-300 · Agentforce Specialist

More certification guides: AI Engineer Certification in 2026: The 7 Real Options, Compared by Stack, AI Product Manager Certification: 5 Real Exams and the Courses Worth Knowing (2026), Agentic AI Certification in 2026: 16 AI Agent Certifications, Grouped by Role and Prompt Engineering Certification in 2026: 10 Real Options, Honestly Compared.

Related guides

• AI Certifications for Beginners: 5 You Can Earn With No Experience
• The AI Cert Starter Kit
• Free AI Certifications in 2026: 15 Free Programs, Honestly Labeled
• AI Exam Prep
• Do AI Certifications Actually Increase Your Salary? An Honest Answer