Blog / CCAR-F vs CCAR-P: Which Claude Certification Should You Take First?
CCAR-F vs CCAR-P: Which Claude Certification Should You Take First?
The names collide on purpose. Anthropic put "Claude Certified Architect" in front of both exams, and left the difference in a single trailing letter: CCAR-F (Foundations) and CCAR-P (Professional). That's easy to skim past when you're picking an exam under deadline pressure, and it leads people to assume CCAR-P is just "CCAR-F but harder." It isn't. They test different things, for different people, at different points in a career. Here's what actually separates them.
The short version
| CCAR-F | CCAR-P | |
|---|---|---|
| Full name | Claude Certified Architect - Foundations | Claude Certified Architect - Professional |
| Fee | $125 | $175 |
| Domains tested | 5 | 7 |
| Typical study time (new to the material) | 30-40 hrs | 45-60 hrs |
| Passing score | 720/1000 | 720/1000 |
Both are $125-$175 exams with the same 12-month validity and free on-time renewal - see the sourced facts for both. The real split is in what they cover, not the logistics.
What CCAR-F actually tests
CCAR-F is about building a single agent well: the request/response loop against the Messages API, designing tools an agent can call reliably (including MCP), configuring Claude Code for real workflows, writing prompts that produce structured, checkable output, and keeping a long-running agent's context from degrading. It's an individual-contributor exam - the unit of concern is one agent, built and operated correctly.
The 5 domains, in this site's CCAR-F study guide:
- Agentic Architecture & Orchestration
- Tool Design & MCP Integration
- Claude Code Configuration & Workflows
- Prompt Engineering & Structured Output
- Context Management & Reliability
What CCAR-P adds
CCAR-P assumes you can already do what CCAR-F tests, then asks a different question: can you be trusted to run this at organizational scale? That means evaluating and tuning agents against real metrics instead of vibes, governance and risk controls that satisfy a security or compliance review, managing an agent through its full lifecycle with non-technical stakeholders in the room, and the operational plumbing (observability, cost control, developer enablement) that a single-agent build doesn't need. It has more domains than CCAR-F and, in this site's own study-hour estimate, meaningfully more recommended prep time - not because the underlying Claude concepts are harder, but because "prove this is safe and observable at scale" is a bigger scope than "build one agent that works."
The 7 domains, in this site's CCAR-P study guide:
- Integration
- Solution Design & Architecture
- Evaluation, Testing & Optimisation
- Governance, Safety & Risk Management
- Stakeholder Communication & Lifecycle Management
- Claude Models, Prompting & Context Engineering
- Developer Productivity & Operational Enablement
So which one first?
If you're not sure, take CCAR-F first. It's the cheaper, narrower exam, and its content - the agentic loop, tool design, prompt engineering, context management - is the foundation CCAR-P builds its governance and evaluation material on top of. Anthropic doesn't enforce a prerequisite between them (no Claude certification requires another), so nothing stops you from attempting CCAR-P cold, but you'd be learning the CCAR-F material anyway, just inside a harder framing.
Go straight to CCAR-P instead if your job is already the CCAR-P job: you're not the one writing the first version of an agent, you're the one deciding whether an agent someone else built is safe to put in front of customers, how it gets monitored once it's live, and how you'd defend that decision to a risk committee. In that case CCAR-F's single-agent framing will feel like it's testing skills you exercise daily as a side effect, and CCAR-P is the exam that actually matches your role.
Undecided which of all four Claude certifications fits you, not just these two? The cert picker asks four questions and points you at one; the comparison page puts all four side by side.