Snowflake Rebuilt Its Certification Program in 2026: Here’s What Actually Changed

Snowflake spent 2026 quietly rebuilding its certification program from the ground up. The flagship SnowPro Core exam got rewritten around AI and the modern data platform, two of the three specialty exams received a version bump within the same week, and a brand-new advanced credential aimed squarely at machine learning operations opened for testing for the first time in the company’s history. If the last time you looked at Snowflake’s certification lineup was even a year ago, several of the exam codes you remember no longer exist in their old form.

Why Snowflake Rewrote the SnowPro Core Exam This Year

Snowflake released COF-C03, the newest version of the SnowPro Core exam, on February 16, 2026, retiring the English-language COF-C02 version on May 14, 2026 (localized COF-C02 translations followed on July 31, 2026). That’s not a routine content refresh. Snowflake’s own fiscal 2026 results show why the timing makes sense: product revenue hit $4.47 billion for the year, up 29 percent, and Cortex Code, the platform’s AI coding agent, had already been adopted by more than 4,400 customers. An entry-level exam built for a data warehouse company doesn’t hold up when the company has become an AI platform company in the space of two years.

What COF-C03 Actually Tests Now

The new blueprint drops entire sections of the old warehouse-fundamentals content and replaces them with material that didn’t exist in COF-C02 at all:

• Snowflake Cortex, tested on positioning and problem-solving rather than API syntax
• Apache Iceberg tables, including how they differ from native Snowflake tables
• Snowflake Notebooks, covering use cases and where they fit rather than hands-on workflows
• Git integration for version-controlled development inside Snowflake
• Governance features such as data lineage, masking policies, row-access policies, and the Trust Center
• Collaboration tools including the Marketplace, Data Exchange, and Data Clean Rooms

Anyone who studied for COF-C02 even eighteen months ago would recognize maybe two-thirds of this exam. The rest is new territory built around how people actually use Snowflake in 2026, not how they used it when the certification was first written.

The Domain Weighting Shift, and Why Architecture Jumped to 31 Percent

COF-C03 also consolidated the exam from six domains down to five, and the weighting change tells you exactly where Snowflake wants candidates to spend their study time:

• Architecture and features, 31 percent
• Performance, querying, and transformation, 21 percent
• Governance, 20 percent
• Loading and connectivity, 18 percent
• Collaboration, 10 percent

Architecture alone jumped from 25 percent under COF-C02 to 31 percent here, and combined with performance it now accounts for more than half the exam. The exam itself is still 100 questions with a 115-minute time limit, costs $175 per attempt through Pearson VUE, and requires a scaled score of 750 out of 1,000 to pass. Snowflake recommends at least six months of hands-on platform experience before attempting it, and a passing result stays valid for two years before recertification.

What Happens If You Already Hold COF-C02

If you’re already certified under COF-C02, nothing changes for your existing badge. You keep your current SnowPro Core status and follow the normal two-year recertification cycle from whenever you originally passed, not a new clock tied to COF-C03’s release. The only real consequence is for anyone starting fresh: COF-C02 simply isn’t bookable anymore, so the current SnowPro Core exam is the only path in for new candidates. Worth noting on DirectCertify’s own catalog specifically: the COF-C02 product listing is still live and selling with no retirement notice on the page itself, even though the exam it prepares you for has been unbookable through Snowflake since mid-May. If you’re shopping for Core prep material, confirm you’re looking at COF-C03 content before buying.

The Specialty Track Quietly Got a Version Bump Too

Snowflake runs three specialty-level certifications sitting above Core: Gen AI, Snowpark, and Native Apps. Two of the three moved to new exam versions on July 24, 2026, the same day, which wasn’t a coincidence so much as a cleanup pass tied to how fast the underlying features had changed since the originals launched. Snowpark’s SPS-C01 wasn’t touched in this cycle and remains the current version.

Gen AI Specialty Moves from GES-C01 to GES-C02

GES-C01 is already flagged “Exam Retired!” on DirectCertify’s own product page, pointing buyers straight to GES-C02 as the replacement. The new version runs 90 questions in 85 minutes and, like the old one, requires an active SnowPro Core certification plus at least a year of hands-on Gen AI experience with Snowflake in a production environment. SnowPro Specialty exams are priced at $225 per attempt, a step up from Core’s $175. Given how much of COF-C03 now leans on Cortex, taking Core first and Gen AI Specialty second is a logical sequence rather than an arbitrary one.

Native Apps Specialty Moves from NAS-C01 to NAS-C02

DirectCertify covered the original NAS-C01 exam when it was still the live version of the Native Apps specialty. That exam has since been retired on Snowflake’s side and replaced by NAS-C02, released the same day as the Gen AI update. The Native Apps track validates building and monetizing applications inside the Snowflake Marketplace using the Native App Framework, and it’s stayed a smaller, more specialized credential than Gen AI even after the refresh, useful mainly for engineers actually shipping apps through Snowflake rather than general data practitioners.

Snowflake Opened Its First MLOps Certification, and It’s Still In Beta

The bigger structural change this year isn’t a version bump at all. Snowflake opened beta testing for SnowPro Advanced: MLOps Engineer, exam code MLA-B01, running from June 15 through July 13, 2026. As of this writing, Snowflake’s own certification listing still shows the exam under that beta-suffixed code, and it isn’t yet available as a product on DirectCertify’s catalog, so don’t expect a practice-test listing for it here until Snowflake formally moves it to general availability. This is the company’s first credential built specifically around production ML lifecycle work rather than general data engineering or platform administration, and it lands at a moment when Cortex and Snowpark ML have pulled a meaningful share of enterprise machine learning workloads directly onto the platform.

Eligibility, Cost, and Exam Format

• At least two years of experience architecting and operationalizing enterprise-scale ML lifecycles in cloud environments
• At least one of those years specifically using Snowflake’s data cloud functionality
• 70 questions, multiple-choice and multiple-select, 115-minute time limit, English only
• $188 per attempt during the beta window, roughly half the $375 a standard SnowPro Advanced exam costs once it’s out of beta, which is a fairly typical beta-pricing discount rather than a permanent price point

What the Five Exam Domains Actually Test

• Operationalizing data preparation and feature engineering, 20 percent
• MLOps infrastructure and management, 24 percent
• Model serving and deployment operations, 18 percent
• Pipeline orchestration, automation, and CI/CD, 22 percent
• Governance, security, and monitoring, 16 percent

Notice that infrastructure management and pipeline orchestration together make up 46 percent of the exam. This isn’t a certification about knowing what a feature store is in the abstract; it’s built for people who are already running ML pipelines in production and need to prove they can keep them governed, monitored, and reproducible at enterprise scale.

Where This Leaves the Rest of the SnowPro Lineup

Between Core, Specialty, and Advanced, Snowflake now runs eleven active SnowPro certifications, according to the current lineup on Snowflake’s own certification page. Most of the Advanced tier wasn’t touched this year at all, so if you already hold or are studying toward one of these, nothing changed underneath you.

We’ve already broken down what the Advanced: Data Engineer credential actually covers if that’s the track you’re weighing against MLOps Engineer, since the two overlap more than their titles suggest. And Snowflake isn’t alone in reshaping a flagship exam around AI this year either; Google went through a similar overhaul of its own certification lineup, which we covered in Google Cloud’s 2026 exam changes, and the pattern of vendors bolting dedicated AI-specialty credentials onto an existing program rather than starting over is becoming the norm across the data platform space, not something unique to Snowflake.

What a Snowflake Certification Is Actually Worth in the Job Market Right Now

Salary Ranges by Experience Level

A 2026 staffing analysis from KORE1, a technology recruiting firm that places Snowflake engineers directly, puts current compensation in three rough bands:

• Mid-level data engineers with two to four years on the platform: $110,000 to $140,000
• Senior engineers with four or more years and an active SnowPro credential: $140,000 to $180,000
• Principal-level architects combining SnowPro Advanced certification with dbt and Snowpark Python depth: $210,000 to $265,000

That last bracket is the important one to understand correctly. Nobody reaches $210,000 on the strength of SnowPro Core alone. It’s the Advanced-tier certifications, stacked on actual production experience with the broader modern data stack, that move the number that high.

Job-Posting Demand and Where It’s Concentrated

On ZipRecruiter’s own listings, hourly pay for roles referencing SnowPro certification spans roughly $20 to $93 an hour, a range wide enough that it captures everything from junior contract analyst work up through senior data engineering positions. That spread lines up with what Snowflake’s growth numbers suggest is happening underneath it: the company added 2,332 net-new customers over the past fiscal year, including 740 in the fourth quarter alone, up 40 percent year over year, and now serves 790 of the Forbes Global 2000. Every one of those net-new accounts eventually needs people who can run the platform, which is the actual demand driver behind the certification, not the badge itself.

Choosing Your Path: Core First, Then Specialty or Advanced

With eleven active certifications now on the board, picking a starting point matters more than it used to. A practical way to sequence it:

1. Start with SnowPro Core (COF-C03) regardless of where you’re ultimately headed. It’s the only certification with no prerequisite, and both the Specialty and Advanced tiers assume you already have it.
2. If your daily work is building or maintaining ML pipelines in production, keep an eye on MLOps Engineer as it moves out of beta rather than defaulting straight to Data Engineer, since the two tracks test genuinely different skills despite the overlap.
3. If you’re shipping features through Cortex or building on the Native App Framework, pick the specialty exam that matches what you actually build day to day, not the one with the more interesting-sounding name.
4. If the goal is the largest possible pay jump, treat an Advanced-tier certification, Architect, Security Engineer, or Data Engineer in particular, as the real destination. Core opens the door; Advanced is what employers are actually paying a premium for.

For anyone still weighing options, DirectCertify’s full lineup of Snowflake practice exams covers every current exam code in the table above except MLOps Engineer, which we’ll add once it clears beta.

One honest note before wrapping up: DirectCertify is an independent certification prep provider and isn’t affiliated with, endorsed by, or sponsored by Snowflake Inc. Exam codes, pricing, and beta timelines are moving fast on Snowflake’s own schedule this year, so confirm the current specifics on Snowflake’s certification page before you register for anything.

Frequently Asked Questions About Snowflake Certification in 2026