Head to head
Cred Farmer vs CCAFPreparation: which should you use?
Cred Farmer scores higher under every reader profile published here, but CCAFPreparation scores higher on evidence and transparency and on value and access.
Cred Farmer publishes this comparison and is one of the two platforms in it. Every figure below links to CCAFPreparation's own page and carries the date we read it. The verdict is computed from published weightings, not written by us.
- The site contradicts itself on how many questions it holds. The homepage and /mock-exam say 408, /practice says "396 questions across the five domains", and the FAQ says "There are 318 items in total, including the 12 official sample questions verbatim." The first two reconcile (396 domain-drill items plus the 12 official samples), and the bank the mock exam ships contains 408, so the FAQ figure is the stale one. Worth noting the outlier is the smallest number rather than an inflated one, but a reader comparing the FAQ against the homepage sees a 90-question gap with no reconciliation offered.
Who wins, by what you need
Each row is that profile's published weighting applied to both platforms' criterion scores. Nothing here is hand-assigned: change the weights and the winners change.
| If you say… | Cred Farmer | CCAFPreparation | Better fit |
|---|---|---|---|
| Balanced overall | 90 | 70 | Cred Farmer |
| I want one system for all four exams | 92 | 69 | Cred Farmer |
| I only need one exam | 92 | 80 | Cred Farmer |
| I want the best free option | 89 | 88 | Cred Farmer |
| I learn from video or an instructor | 52 | 47 | Cred Farmer |
| My exam is in two weeks | 96 | 78 | Cred Farmer |
| I am preparing a team | 82 | 46 | Cred Farmer |
What each one is better at
Where CCAFPreparation is stronger
- Every substantive claim in a lesson's Deep Dive links to the Anthropic page it came from and the date that page was read, and a build-time check fails the site rather than publishing an unsourced fact.
- A dedicated divergence log records each point where the July 2026 exam guide and current Anthropic documentation disagree, gives both versions, and marks which answer to give in the exam.
- Every option in every question carries its own explanation, not only the correct one, and the practice drills break down to per-task-statement counts: Domain 1 publishes 13 questions against each of its seven task statements.
- The mock offers a scenario-grouped mode that keeps each scenario's items together, so you rehearse holding one production context across a run of questions instead of switching every item.
- It is explicit about the limits of its own numbers: the terms and the mock page both say the scaled score is an approximation because the real scaling function is not published, and state that the live item bank is confidential and nobody there has seen it.
- The whole site is genuinely free with no account, no email capture and no cookies, and the terms confirm the footer tip unlocks nothing.
- Evidence and transparency: scores 5/5 against our 4/5.
- Value and access: scores 5/5 against our 4/5.
Where Cred Farmer is stronger
- Certification coverage: we score 5/5 against their 2/5.
- Team and operational readiness: we score 3/5 against their 0/5.
- Diagnostics and personalisation: we score 5/5 against their 3/5.
- Practice depth: we score 5/5 against their 4/5.
- Mock-exam realism: we score 5/5 against their 4/5.
Feature by feature
Expand any cell to see the words we read and where we read them. If a link no longer says what we quote, the figure is out of date and we want to know.
| Cred Farmer | CCAFPreparation | |
|---|---|---|
| Certifications covered | All four: CCAO-F, CCDV-F, CCAR-F and CCAR-P, in one account, each with five timed forms, drills by objective, a readiness score, a study guide mapped to the published domains and a flashcard deck.Evidence“Study guides mapped to every domain, spaced-repetition flashcards, and five timed forms per exam built to its published blueprint are available across all four tracks, and every timed form contributes blueprint-based practice evidence.” | CCAR-F only (Claude Certified Architect – Foundations), built against exam guide v1.0 (July 2026). There is no curriculum, question bank, flashcard deck or mock for CCAO-F, CCDV-F or CCAR-P; those codes appear only in two blog posts comparing the four certifications.Evidence“Exam code CCAR-F · Guide v1.0 (July 2026)” |
| Practice questions | 1,180, every option explained. 300 CCAO-F, 265 CCDV-F, 300 CCAR-F and 315 CCAR-P across the four gated banks, every item tagged to a published domain and objective and practised in the timed forms, in drills by objective or by domain, and in drills built from your own misses. A separate public set feeds the no-account practice papers and samples.Evidence“1,180 questions written to the published exam blueprints, five timed forms per exam at the real exam's length and time limit, and a reason for every option, right or wrong, so you learn from each mistake.” | 408 in the full bank. The /mock-exam per-domain bank table sums to 408 (93+67+83+85+80), and counting the items the mock-exam page actually ships gives exactly 408 distinct questions. 396 of them are reachable through the per-domain practice drills (91+65+79+83+78); the remaining 12 are the official sample questions from the exam guide, reproduced verbatim. The FAQ still states a lower figure of 318, which the site's own bank contradicts.Evidence“Questions are drawn from a bank of 408, sampled per domain in the official blueprint proportions rather than evenly” |
| Study material | Exam-focused study guides. One per certification, organised by the published domains and objectives, with written hands-on labs, a glossary and per-domain cheat sheets, and a public study guide per certification that reads with no account. Written material only: no video and no instructor.Evidence“Focused Study Guides Structured chapters mapped directly to the exam objectives.” | 30 lessons, one per official task statement, ordered and weighted to the blueprint. Each lesson carries the exam-guide objectives, a written explanation, a Deep Dive whose every section ends in a dated source line, exam traps, a quick-reference table, an inline quiz, a build exercise with a walkthrough and a graded hint ladder, and an appendix of copy-paste study prompts to run in Claude, one of which ends in a green/amber/red verdict per concept. Alongside them: 5 printable per-domain cheat sheets, a 219-term searchable glossary, a summarised exam-guide section, 6 per-scenario pages, a divergence log recording where the July 2026 exam guide and current Anthropic documentation disagree, and 7 long-form exam-notes articles.Evidence“One lesson per official task statement — 30 in total, weighted exactly as the exam blueprint weights them.” |
| Full-length timed mocks | 20 full-length timed forms: five per certification at the real item count and 120-minute clock (60 / 53 / 60 / 63 items), each built to its published domain weights, four as mock exams and the fifth an objective-mastery form. The CCAR-F forms each present four of the six published scenarios with their items grouped under them. Flag items as you go, then review the reason for every option by domain and by objective.Evidence“In length, they match: every timed form in your account has the real exam's question count and time limit (CCAO-F 60 questions in 120 minutes, CCAR-F 60 questions in 120 minutes, CCAR-P 63 questions in 120 minutes and CCDV-F 53 questions in 120 minutes).” | One full-length timed mock, offered in two modes. Standard: 60 questions in 120 minutes drawn per domain in blueprint proportions (16/11/12/12/9), flaggable, auto-submitting when the clock runs out, scored on the real 100 to 1000 scale against the 720 pass mark with a per-domain breakdown. The second mode, 'Start grouped by scenario', keeps each scenario's items together to mirror how the real sitting frames a run of questions. Each sitting redraws from the 408-item bank, and no attempt limit is stated. The site says its scaled score is a linear approximation because the real scaling function is not published.Evidence“60 questions · 120 minutes · pass mark 720 of 1000. Questions are drawn per domain in the blueprint proportions, so the section you are weakest in is the section that costs you the most.” |
| Flashcards | Yes: a rapid-recall deck per certification inside the account, served through a spaced-repetition drill, plus a 40-card open deck per certification with no account.Evidence“Flashcard drills resurface each concept right before you'd forget it.” | 385 cards in five decks, one per domain: 80 / 78 / 72 / 80 / 75. Each deck is also published as a plain full list for scanning or printing.Evidence“CCAR-F Flashcards — 385 Cards Across All 5 Exam Domains — CCAF Preparation” |
| Spaced repetition | Yes: an SM-2 scheduler, scheduling each card individually from a four-way rating, with lapses requeued inside the same session.Evidence“Study guides mapped to every domain, spaced-repetition flashcards, and five timed forms per exam built to its published blueprint are available across all four tracks, and every timed form contributes blueprint-based practice evidence.” | Yes: a box-based scheduler on the flashcard decks. A card marked 'Again' returns at the end of the round and again the next day; a card you keep getting right moves up a box and is suppressed for progressively longer, capped at about a fortnight. The site describes the mechanism in its own words rather than naming an algorithm, and the scheduling state is stored in the browser only.Evidence“Anything you mark “Again” comes back at the end of the round, and again tomorrow — a card you keep getting right moves up a box and goes quiet for longer each time, up to a fortnight.” |
| Diagnostics | Objective-level: every timed form is scored by domain and by objective, the weakest domains are ranked from measured results and linked to the chapter that covers them, and each objective you got wrong routes you to a drill on it. Drills can also be built by domain or from your own misses.Evidence“Beside the score, each objective you got wrong routes you to a drill on that objective.” | No standalone diagnostic or placement quiz. The prescribed diagnostic is sitting the full 60-question, 120-minute mock cold and reading the per-domain breakdown, which the mock page calls the figure worth acting on. Separately, /practice/review pools every question you have missed across lesson quizzes, practice scenarios and build walkthroughs and drills them on their own, dropping an item once you answer it correctly.Evidence“Sit a mock exam cold. Before reading anything. A score you have not studied for is the only honest map of where your gaps are” |
| Readiness / adaptive path | A per-certification readiness score gated on measured answers: 60% performance, 40% preparation (study path, objective coverage and streak). The first sitting of each form, guided or timed, sets the band and preparation moves the score only within it: below the pass mark it stays under 60, and Exam-ready (90) needs a first-sitting average of 800. Drills, re-sits and the diagnostic never count as evidence, and a study section counts only when its checkpoint questions are answered correctly. Set an exam date and the hours you can give, and a dated plan lays the work out day by day: sections to read, drills on objectives measured weak, and full-length mocks spread across the days before the exam. It says when the work does not fit, and it makes no pass prediction.Evidence“Study path progress, objective coverage and your streak move the score within that band, never past it.” | A readiness score on /progress blends lesson completion, practice accuracy and mock performance, weighted by how much evidence each provides, so ticking a lesson counts for less than answering questions, but it still counts. It is not adaptive question serving: practice and mock draws are fixed blueprint proportions and are never reweighted toward measured weak domains.Evidence“Readiness blends lesson completion, practice-question accuracy, and mock-exam performance, weighted by how much evidence exists for each.” |
| Free tier | All free during open beta: all four certifications, every study guide, every timed form and every drill. With no account: a timed practice paper, six sample items, a 40-card deck and a study guide per certification; the rest needs a free account.Evidence“Right now, Cred Farmer is completely free while we are in open beta.” | Everything is free, and there is no tier structure at all: no paid plan exists, so nothing is gated. Every lesson, all practice questions, the full mock, the flashcards, the cheat sheets and the glossary are open without a login.Evidence“Every lesson, every practice question, the full mock exam, the flashcards, the cheat sheets and the glossary are open with no login, no paywall and no locked content.” |
| Price | $0 while in open beta. No paid tier is published and no pricing is announced for after the beta.Evidence“Individual preparation is free while in open beta.” | Nothing is sold and no price exists. A voluntary tip link sits in the footer through a third-party payment provider; the terms state it is not a purchase and that nothing is gated behind it or unlocked by it.Evidence“The support link is a voluntary tip through a third-party payment provider.” |
| Account required | Yes for the full study guides, the five timed forms per exam, the drills, the readiness score and the dated plan: a free account via Google or GitHub sign-in. The practice paper at /practice/<code>, the sample items, the open flashcard decks, the public study guide on each certification page, the blog and this comparison need no account.Evidence“To maximize security, we only use OAuth (Google and GitHub).” | No, and there is no account to create. No sign-in, no password handling, no email capture and no cookies; progress lives in three browser localStorage keys and is never transmitted. The privacy page names cookieless Vercel Web Analytics and Speed Insights, which set no cookie and collect only aggregate page-view and performance data, as the only measurement on the site.Evidence“No user accounts, sign-in, or password handling” |
| Team support | An organisation layer, in pilot and set up on request: a cohort dashboard showing each consenting member's readiness score, gate state and mock count, member and invitation management, and recorded exam-fee sponsorship offers against a readiness bar the organisation sets. No published seat pricing, no self-serve creation and no procurement route yet.Evidence“Set the readiness score you are willing to fund at, see who has crossed it, and record an offer against a certification.” | Not published |
| States independence from Anthropic | Yes: stated in the footer of every page and repeated on the certification pages.Evidence“Not affiliated with, or endorsed by, Anthropic or Pearson VUE.” | Stated in the site-wide footer on every page and restated on /about, /terms and in the FAQ, with trademark attribution. The terms add that nothing read there grants any credential and direct exam registration to the certification programme itself. No endorsement or affiliation is claimed anywhere.Evidence“Independent study resource. Not affiliated with, endorsed by, or sponsored by Anthropic, PBC. Claude and Claude Certified Architect are trademarks of Anthropic, PBC.” |
Which to choose
Choose CCAFPreparation if…
- You are sitting CCAR-F and want to be drilling within thirty seconds of landing: everything, including the full 60-question mock, opens with no login, no email and no cookies. Cred Farmer's no-account CCAR-F set is a 20-item quick mock; its full-length forms sit behind a free account.
- You want to check where a claim came from before you trust it. Every Deep Dive statement links to the dated source behind it, and the divergence log tells you what to answer when the July 2026 exam guide and current documentation disagree: a level of provenance we do not publish per claim.
- CCAR-F is the only exam you care about and you want the deepest single-exam bank you can get for nothing: 408 items for that one certification, against roughly 300 in ours, with the 12 official sample questions reproduced verbatim.
- You want to rehearse the scenario framing with no account, and pick how the paper is laid out. Their one mock runs as a standard blueprint-proportion draw or grouped by scenario, with no login. Every Cred Farmer CCAR-F form presents four of the six published scenarios with its items grouped under them, but it sits behind a free account.
- You would rather nothing about your study leave your machine. Their progress and flashcard scheduling never reach a server; ours are stored against an account.
Choose Cred Farmer if…
- You want to practise the way the exam is sat: five full-length timed forms per exam at the real item count and clock, with a reason for every option when you review.
- You want each study session chosen by what you got wrong: every missed objective routes you to a drill on it, and you can drill by domain or from your own misses.
- You have booked an exam and want an honest read of your practice results, from a score capped by your first sitting of each form rather than by activity, and a dated plan that lays out each day until the exam.
- You want to sit a free timed practice paper for your exam right now, with no account, and see a raw score by domain.
- You expect to sit more than one Claude certification and want one setup for all four.
- You are preparing other people and want to see who is measurably ready, with each person opting in, before you fund an exam.
What CCAFPreparation does not offer
- Single-exam scope: CCAR-F only. A candidate sitting CCAO-F, CCDV-F or CCAR-P gets no curriculum, question bank, flashcard deck or mock here: only two blog posts comparing the four certification codes.
- No account and no server-side storage, so progress, readiness, the missed-question list and flashcard scheduling live in one browser's localStorage. Their own page says: "This list lives in your browser only. Clearing site data or switching device starts it over." There is no cross-device sync and no recovery.
- No team, seat, cohort or admin capability of any kind: no roster, no manager-visible progress, no reporting and no multi-seat purchasing.
- No short diagnostic or placement instrument. The only measurement path is the full 60-question, 120-minute mock, which the site tells you to sit cold before studying.
- Question draws are fixed blueprint proportions in both practice and the mock, with no adaptive selection weighted toward your measured weak domains; the only targeted drill is the manually accumulated missed-question list.
- The mock's published output is a scaled score and a per-domain breakdown; no item-by-item review of a completed paper is described, and there is one mock surface redrawn from the bank rather than a set of distinct published papers.
Observable gaps in what the site publishes, not judgements about quality. We have run no blind review of anyone's questions, including our own, so nothing here compares how good the content is, only what each platform offers.
Sources
- https://www.ccafpreparation.com/ Homepage: exam code and guide version, and the instruction to sit a mock cold as the diagnostic
- https://www.ccafpreparation.com/mock-exam Mock format (60 questions, 120 minutes, 720 pass mark), the per-domain bank table totalling 408, the scenario-grouped mode and the scoring caveat
- https://www.ccafpreparation.com/learn Curriculum structure: 30 lessons, one per official task statement, weighted to the blueprint
- https://www.ccafpreparation.com/practice Per-domain drills totalling 396 questions, each with an explanation for every option, framed as what you drill after the mock
- https://www.ccafpreparation.com/practice/review The missed-question pool, and the statement that the list lives in the browser only
- https://www.ccafpreparation.com/flashcards 385 cards across five per-domain decks
- https://www.ccafpreparation.com/flashcards/agentic-architecture The box-based review scheduler and its published behaviour
- https://www.ccafpreparation.com/progress The readiness score and what it blends
- https://www.ccafpreparation.com/faq Free-tier scope, and the stale 318-item question count
- https://www.ccafpreparation.com/about Independence disclaimer, per-claim sourcing and the build gate, the divergence-log policy, and the corrections invitation
- https://www.ccafpreparation.com/terms The voluntary tip is not a purchase; no-guarantee and confidential-item-bank statements; last updated 13 August 2026
- https://www.ccafpreparation.com/privacy No accounts, sign-in or password handling; the three localStorage keys; cookieless Vercel analytics
All read on 3 September 2026. Found something wrong? Email hello@credfarmer.com and it will be corrected.
Common questions
Is CCAFPreparation better than Cred Farmer?
Cred Farmer scores higher under every reader profile published here, but CCAFPreparation scores higher on evidence and transparency and on value and access. Under the balanced weighting, CCAFPreparation scores 70 and Cred Farmer scores 90 out of 100. The weightings are published and adjustable, so the honest answer depends on which exam you are sitting and what you can spend.
Is this comparison independent?
No. Cred Farmer publishes it and is one of the two platforms in it. That is why the rubric is published, every figure links to CCAFPreparation's own pages with the date it was checked, and the verdict above is computed from those published weightings rather than written by us.
Is CCAFPreparation an official Anthropic product?
No. Neither is Cred Farmer. Anthropic publishes the exam guides and Pearson VUE delivers the exams; every platform in this comparison is independent of both.
How current are these figures?
Everything here was read on CCAFPreparation's own site on 3 September 2026. Prices, question counts and free tiers change without notice, so follow the source links before you rely on a figure, and email hello@credfarmer.com if you find one out of date.
Other comparisons
Not affiliated with, or endorsed by, Anthropic or Pearson VUE.