Home › Certifications › How to Pass CCAR-F
Claude Certification Guide · 2026How to Pass CCAR-F in 2026: The Exam Prep Guide
Not what the credential is — how to actually pass it. The published six-scenario bank and why only four reach your sitting, how to split study hours in blueprint proportion instead of evenly, what a scaled 720 really demands, the five ways candidates fail, and a 6-week plan built around drills rather than reading.
Authorized Training & Technology Partners
Claude
Claude Certified Architect – Foundations (CCAR-F)
The exam this guide prepares you forProctored delivery — online or at a test centre
Registration via the Anthropic Partner AcademyPassing CCAR-F comes down to one structural fact most candidates never exploit: Anthropic publishes all six exam scenarios, and your sitting draws only four of them at random. The syllabus is not a mystery — it is a reading list. Work all six until you can defend an architecture for each, allocate study time in blueprint proportion rather than evenly, and the 720 becomes arithmetic instead of luck.
This guide is deliberately about execution. If you want the credential explained — what tier it sits at, what it signals to employers, how it compares with the associate and developer tracks — that is our CCAR-F certification guide. If you want it read as one half of an agentic engineering skillset, that is the agentic architecture guide. What follows assumes you have already decided to sit the exam and want to pass it first time.
The six-scenario bank is the whole exam
Most certifications hide their content behind a blueprint of abstractions. CCAR-F does something unusual: it publishes the scenarios themselves. Six of them exist, and any given sitting presents four, chosen at random. That single design choice should reshape how you study.
Your odds of meeting any one scenario
Four scenarios drawn at random from a published bank of six
Calculated by us from the published format. Four scenarios are drawn from six, so there are 15 possible combinations. Skip one scenario and only 5 of those 15 combinations avoid it — a 1 in 3 chance. Skip two and just 1 of 15 avoids both, about 6.7%. The maths is unforgiving: there is no rational reason to skip a scenario.
The failure mode we see most often: candidates read all six scenarios, recognise them on exam day, and still lose marks. Recognition is not the skill being tested. Each scenario asks you to choose an architecture and defend the trade-off. If you have not written down — or better, built — your answer for each one before the exam, you will be designing under a two-minute clock for the first time. Build all six. Then argue with yourself about each.
Study in blueprint proportion, not evenly
The second-biggest lever is where your hours go. CCAR-F publishes exact domain weightings, so an even split across five domains is a measurable mistake — it overspends on the smallest domain and underspends on the largest by roughly 12 percentage points.
Where your study hours should go
CCAR-F domain weightings, and the hours each earns in a 60-hour plan
Source: Anthropic Partner Academy, CCAR-F exam blueprint; hours are the weightings applied to a 60-hour plan, rounded. Against an even split of 12 hours each, proportional study moves 4 hours into orchestration and 3 out of context management. On 60 questions that is the difference between comfortably clearing 720 and missing it.
What a scaled 720 actually demands
CCAR-F scores on a scaled 100–1000 range with 720 to pass. Scaled scoring means 720 is not simply 72% of questions correct — items are weighted by difficulty, so the raw percentage you need moves between forms. Two practical consequences follow.
First, do not aim for 72%. Aim for a comfortable margin, because you cannot compute your own raw threshold during the exam. Second, because the exam is scenario-based rather than recall-based, the marks cluster: misreading a scenario's core constraint costs you several linked questions at once, not one. Errors on this exam are correlated. That is why scenario preparation outperforms flashcards.
The five ways candidates fail
Over-engineering the answer
Given a scenario, candidates reach for the most sophisticated architecture — multi-agent where one agent suffices, an orchestration framework where a bounded loop would do. The simpler design that meets the stated constraint is usually the correct option. Sophistication is not the thing being assessed; judgement is.
The most common single errorMissing the constraint sentence
Scenario stems are long, and the sentence that decides the answer is often the last one — latency, cost ceiling, operational simplicity, compliance. Several options will be workable architectures. Only one serves the constraint. Read the final sentence first, then the stem.
Read backwardsNever having used Claude Code
20% of the exam is Claude Code configuration and workflows — permissions, project setup, integration into a development process. It is examined as a working tool, and it is the domain that most obviously separates candidates who have used it daily from those who have read about it. You cannot revise your way through this one.
A fifth of the marksMiscounting multiple-response items
Multiple-response questions state how many answers to select. Under a two-minute clock candidates select two when the item asked for three, and lose a question they actually knew. This is pure, avoidable loss. Rehearse the format until checking the count is automatic.
Free marks, thrown awayTreating context management as an afterthought
At 15% it is the smallest domain, so it gets studied last and least. But context strategy — what stays in the window, what gets summarised, how degradation is detected — is where candidates without production experience are most exposed, because the questions assume you have watched an agent lose coherence at length.
Smallest domain, biggest gapWhat passing actually looks like
Candidates who pass comfortably tend to share one habit: they built things and broke them on purpose. An agent that loops forever, a tool suite the model cannot choose between, a context window that silently degrades. Every scenario question is a description of a failure someone has already had — having had it yourself is the shortcut.
The pattern behind the passesCCAR-F exam mechanics at a glance
| Attribute | Detail |
|---|---|
| Credential | Claude Certified Architect – Foundations (CCAR-F) |
| Issued by | Anthropic |
| Format | 60 scenario-based items — multiple-choice and multiple-response, each stating how many to select |
| Scenarios | 4 drawn at random from a published bank of 6 |
| Duration | 120 minutes — 2 minutes per question |
| Scoring | Scaled 100–1000; 720 to pass (not a flat 72% of items) |
| Domains | Orchestration 27% · Claude Code 20% · Prompting 20% · Tools & MCP 18% · Context 15% |
| Exam fee | US$125 (checkout price reflects any applicable partner-tier discount) |
| Delivery | Pearson VUE — online proctored or at a test centre |
| Registration | Anthropic Partner Academy, via the Claude Partner Network (free at Registered level) |
| Validity | 12 months from the date awarded |
| Note on naming | Early material abbreviated this credential CCA-F; the current Pearson VUE exam code is CCAR-F |
What passing is worth
The honest case for CCAR-F is not a pay multiple — it is scarcity. The credential is new enough that very few people hold it, and it is issued by the lab that builds the model, which is a different kind of signal from a platform badge.
The AI architect progression
Entry
Mid
Senior
Architect
AI architect pay by market
India and the US shown as two separate charts, because rupee and dollar bands are different measures on different scales and should never share an axis.
India — annual CTC
AI engineer and architect roles, ₹ lakh per annum
Bar length maps the upper bound of each band against a ₹53 L scale. Bands compiled from published Indian AI engineer and AI architect ranges.
United States — annual base
AI engineer and agentic specialist roles, US$ thousands
Bar length maps each figure against a $200K scale. Base pay only, excluding bonus and equity. The spread is wide because the specialism is young and titles are inconsistent.
These are role bands, not certification outcomes. No published figure attaches a salary increase to holding CCAR-F — the credential is too new for that data to exist. The scenario-combination arithmetic above is ours, derived from the published 4-of-6 format; it is a statement about exam coverage, not about pay.
Compensation figures are compiled from independent, publicly available industry sources and are shown for role context. They are not a guarantee of pay in any specific market, company or outcome, and 360DT does not promise a salary result from any certification or programme.
Your 6-week CCAR-F prep plan
Weighted to the blueprint, ending on drills rather than reading. Assume 8–10 hours per week, roughly 60 hours total, and build with Claude throughout — a scenario exam punishes reading-only preparation harder than most.
Read all six scenarios, then orchestration
Start with the bank — every scenario, before anything else, so the syllabus arrives as an answer to problems you have already met. Then open the largest domain: agentic loops, stop conditions, step limits, and error handling mid-run. Build one loop and break it deliberately.
Multi-agent systems and enforcement
Supervisor, pipeline and peer-delegation topologies. Hand-off and shared state. Hooks and enforcement. And the highest-scoring judgement on the exam: when one agent is simply better than three. Completes the 27% domain.
Tool design and MCP
Boundaries, naming, descriptions, parameter design. Build a suite of tools, then make two overlap and watch the model hesitate. Then MCP end to end: servers, transports, capabilities. Connect one server to two clients to see what the standard actually buys you. Completes the 18% domain.
Claude Code, used daily
Project configuration, permissions and what the agent may touch, workflow design, and integration into a real development process — including CI/CD. Use it for actual work this week. A fifth of the exam assumes you have, and it is the domain you cannot revise into existence.
Prompting, structured output, context
System prompt design and few-shot construction, then schema-constrained output and what to do when it still comes back malformed. Then context: window strategy, summarisation, memory stores, and detecting long-run degradation before a user does. Completes the last two domains, 35% between them.
Scenario drills and timed mocks
Return to all six scenarios and, for each, write the architecture you would propose and the trade-off you would defend. Then full 60-question, 120-minute mocks scored by domain. Rehearse the multiple-response count until checking it is automatic. Then book.
Two minutes per question, and scenario stems are long. Read the final sentence of the stem first — it usually names the constraint being optimised for, and that single word decides between two otherwise workable architectures. Flag anything over three minutes and return to it. And when two designs both satisfy the constraint, pick the simpler one.
Learn from Anthropic Authorized Instructors
Certification prep at 360DT is delivered live by AWS, Microsoft, Cisco and NVIDIA certified trainers and Anthropic Authorized Instructors — our full international faculty.
Bipeen
AWS, Microsoft, and Anthropic Certified Trainer
AI, machine learning, and cloud transformation expert with more than 25 years of experience. A renowned global corporate trainer and conference speaker known for practical, outcome driven sessions.
Florian Garcia Compte
Cisco Certified Systems Instructor (CCSI)
Partner Director at AVAANZA FORMACION, a Cisco Learning Partner based in Madrid. Certified Cisco Systems Instructor (CCSI No. 21053) specializing in Cisco Data Center (ACI/SDN), Nexus, and Wireless, delivering official Cisco certification courses including CCNA, CCNP Enterprise, and CCNP Data Center.
Shantanu Pandey
Microsoft, Google, NVIDIA and HPE Certified Trainer
Director of Engineering at Meteoros Automation, ranked in Microsoft's Top 100 Trainers of 2025, holding MCT, Google Cloud, NVIDIA and PeopleCert credentials, with 500+ trainings delivered since 2011.
Vikas Mittal
Google and Anthropic Authorized AI Instructor
Veteran technology trainer with more than 25 years of rich industry experience spanning Google, Microsoft, and emerging AI platforms. An exceptional mentor recognized for proven teaching excellence.
Deep
Anthropic Authorized Instructor and AI Cloud Specialist
Seasoned technology leader with over 20 years of experience across AI, cloud computing, and enterprise training. Known for simplifying complex ideas and delivering practical, real world learning.
Sid
Anthropic Authorized Instructor and Azure Data Engineer
Dynamic corporate trainer and consultant with more than 10 years of experience across Microsoft Fabric, Azure Data Engineering, and modern cloud platforms. Focused on clear, practical learning that sticks.
Arshad Ahmad
Microsoft Certified Trainer (MCT), Cybersecurity & Power Platform Expert
Technology Trainer with 15 years in strategic client delivery, holding MCT, Azure Solutions Architect Expert, Cybersecurity Architect Expert and Power BI Data Analyst credentials, with 200+ trainings delivered.
Sravia
NVIDIA Authorized Instructor and AI Data Specialist
Passionate AI and data evangelist with over 10 years of experience across machine learning, MLOps, and cloud native technologies. Dedicated to delivering impactful, engaging enterprise training programs.
Johan
Microsoft Business Applications MVP and Microsoft Certified Trainer
D365 Customer Service Lead – Europe at Avanade, and founder of the Power Platform School. A Microsoft Business Applications MVP and Microsoft Certified Trainer based in London, specializing in Dynamics 365 Customer Service and Power Platform.
Akim Nyamande
Microsoft, CompTIA, Juniper and Cisco Certified Trainer
IT training facilitator with over five years of experience and certifications across Microsoft, CompTIA, Juniper, and Cisco. Previously a Network Administrator before moving into technical training, now delivering hands-on courses in networking, systems administration, and cybersecurity fundamentals.
Ali El Khatib
Cisco Certified Systems Instructor (CCSI) and Microsoft Certified Trainer
Infrastructure Engineer at RHUH with 13 years in the training field, holding CCSI, MCSE, CCNP Routing and Switching, CCNP Security, CompTIA and Microsoft Azure certifications.
Prepare for CCAR-F live, in blueprint proportion
The 360DT Claude Certified Architect Prep — Foundations course covers all five domains weighted exactly like the exam, and builds all six official scenarios with you in class — so whichever four your sitting draws, you have already solved it once. Live online, Sat–Sun, 8:00–11:00 PM IST, delivered by Anthropic Authorized Instructors.
Explore the CCAR-F Prep Course
CCAR-F exam preparation frequently asked questions
Are the CCAR-F exam scenarios really published?
Yes. Anthropic publishes a bank of six scenarios, and any given sitting presents four of them at random. This is unusual and deliberate: it shifts the exam away from recall and toward whether you can reason inside a situation you have already read. It also means there is no rational reason to skip one — skipping a single scenario leaves only a 1 in 3 chance your sitting avoids it.
Is 720 the same as scoring 72%?
No. CCAR-F scores on a scaled 100–1000 range, so items are weighted and the raw percentage needed shifts between exam forms. Aim for a clear margin rather than calculating a threshold. Because the exam is scenario-based, errors also cluster — misreading one scenario's constraint can cost several linked questions at once.
How long should I study for CCAR-F?
Around 60 hours over six weeks for someone already building with Claude — the plan on this page, allocated in blueprint proportion rather than evenly. If you have never used Claude Code, add time: it is 20% of the exam and cannot be learned from reading.
What is the most common reason people fail?
Over-engineering. Candidates reach for multi-agent designs and orchestration frameworks where a single bounded loop meets the stated constraint. The simpler architecture that satisfies the requirement is usually correct. Close behind: missing the constraint sentence at the end of a long scenario stem, and never having actually used Claude Code.
Do I need to build the scenarios, or is reading them enough?
Reading gets you recognition; the exam tests judgement. For each scenario you should be able to state the architecture you would propose and defend the trade-off, under time pressure. Writing that down for all six is the minimum. Building them is better — which is exactly what our prep course does in class.
Is it CCAR-F, CCA-F or CCFA?
The credential is Claude Certified Architect – Foundations. Early material abbreviated it CCA-F; the current Pearson VUE exam code, and what appears on the credential itself, is CCAR-F. CCFA is not an official code — it appears in some third-party and legacy listings but does not come from Anthropic. Use CCAR-F.
How is this different from your other CCAR-F guides?
This one is about passing the exam. Our CCAR-F certification guide explains the credential itself — tier, positioning, career value — and the agentic architecture guide reads it as the architecture half of an agentic engineering skillset alongside AI-103.
Should I sit CCAR-F or go straight to CCAR-P?
Foundations first unless you are already architecting Claude systems at scale. CCAR-P is the professional tier and assumes the Foundations material rather than re-teaching it.
Sources and further reading
- Anthropic Partner Academy — Claude Certified Architect – Foundations exam blueprint (five domains with exact weightings, 60 scenario-based items, 120 minutes, scaled 100–1000 with a 720 pass mark, six-scenario bank with four presented per sitting, US$125 fee, 12-month validity)
- Pearson VUE — Claude Certification Program by Anthropic, used for the exam code and delivery details
- Anthropic Academy — Claude Certified Architect – Foundations certification page, used for the credential's formal name
- Scenario-combination arithmetic (15 possible draws of 4 from 6, and the 1-in-3 and 6.7% coverage figures) was calculated by us from the published exam format
- Published Indian and US AI engineer, AI architect and agentic specialist compensation ranges, used for role context only
360DT is an independent training provider. The CCAR-F exam is administered independently by Anthropic through Pearson VUE and is not included in programme tuition. Exam details are accurate as of 21 August 2026; always confirm current format, pricing, validity and blueprint version with Anthropic before booking.