06-reference/research

snowpro genai specialty c02 exam structure

2026-06-18·research-brief·source: deep-research
snowprogenaicertificationphdatasnowflake

Verifying the SnowPro Gen AI Specialty (GES-C02) Exam Structure: Domain Count, Weights, and Mechanics

The question

"From the auth-gated official SnowPro GenAI Specialty (C02) study guide, confirm the exact domain count and weights, whether document processing is a standalone weighted domain or folded into Functions, and the current question count / time limit / passing score." Context: this verifies the exam-structure facts in the 2026-06-17 study-path brief — the founder's MAIN bet, target exam 2026-08-24, one of two phData +$5k cert escalators.

What we already know (from the vault)

What the web says

The official Snowflake C02 page exposes only the fee and launch date; the structured exam mechanics live in the auth-gated study guide PDF, so all per-figure detail below comes from public exam-overview/practitioner pages reporting on the inherited C01 structure. Confidence flagged per figure.

Figure Value Source / confidence
Exam fee $375 / attempt learn.snowflake.com (official) — HIGH
C02 launch date 2026-05-19, replacing GES-C01 learn.snowflake.com (official) — HIGH
Domain count (C01 baseline) 4 domains Snowflake Builders blog + quick-facts pages — MEDIUM (C01, not confirmed for C02)
Domain 1: Snowflake for Gen AI Overview 26% multi-source agreement — MEDIUM (C01)
Domain 2: Gen AI & LLM Functions 40% multi-source agreement — MEDIUM (C01)
Domain 3: Gen AI Governance 22% multi-source agreement — MEDIUM (C01)
Domain 4: Document AI / document processing 12% multi-source agreement — MEDIUM (C01)
Feb-2026 rebalance (3-domain) 30 / 44 / 26, Document AI removed Sharma practical-prep post — LOW-MEDIUM (third-party, transitional)
Question count ~55 questions quick-facts + community pass accounts — LOW-MEDIUM (C01-inherited, third-party)
Time limit 85 minutes quick-facts + community pass accounts — LOW-MEDIUM (C01-inherited, third-party)
Passing score 750 / 1000 (scaled) quick-facts + community pass accounts — LOW-MEDIUM (C01-inherited, third-party)
Question types multiple-choice, multiple-select, some interactive scenarios community — LOW-MEDIUM

Convergences and contradictions

Synthesis for RDCO

The verification result is reassuring and slightly anticlimactic in the best way: the 2026-06-17 study-path brief does not need a structural correction. Its exam-mechanics figures (55q / 85min / 750-scaled / $375) are confirmed to the extent public sources allow, and its honesty about C02 weights being unpublished proxies is exactly right. The founder can study against that brief with confidence that nothing in its scaffolding is wrong — the only thing still genuinely unknown (standalone-vs-folded document processing) was already named as the open item, and this pass confirms it cannot be closed without logging into the portal. The single concrete upgrade: the $375 fee and 2026-05-19 launch are now official-source-confirmed rather than third-party.

For study-time allocation, the verified relative emphasis is the load-bearing takeaway and it is robust across both the C01 (26/40/22/12) and the Feb-2026 (30/44/26) shapes: Functions is decisively the largest domain at ~40-44%, Overview second at ~26-30%, Governance third at ~22-26%, and document processing the lightest (12% standalone under C01, absorbed under the 3-domain shape). The founder should pour hours into the Cortex/LLM function catalog and the RAG/agent assembly, treat document-processing functions as part of that Functions block rather than a separate heavyweight unit, and — per every pass account — over-index on the RBAC/privileges and cost-billing traps inside Governance where data-engineers reliably lose points. None of this changes from the 2026-06-17 plan; this brief simply removes any doubt that the plan's foundation is sound.

The one action this verification reinforces: the auth-gated study guide is the only source that can confirm C02's actual domain count, per-domain weights, and current mechanics. That is a 10-minute founder task (log into cp.certmetrics.com/snowflake, download SnowProGenAIStudyGuideC02.pdf) and it would convert every MEDIUM/LOW figure above to HIGH. Until then, planning to the safest shape (3-domain 30/44/26, document processing folded into Functions) over-prepares the lightest area at negligible cost — the correct default.

Open follow-ups

Related

Sources