CCDV-F vs CCAR-F: Domain Weights and Format Compared

The question of ccdv-f vs ccar-f comes up early in every Claude Partner Network study plan. Both exams carry the same $125 price tag, the same 120-minute window, and the same 720 passing threshold on a 100 to 1000 scale. They differ in almost everything else: question count, domain structure, target role, and the level of abstraction at which Claude is tested. Picking the right exam first accelerates your study timeline and career positioning. This guide compares both credentials across every dimension that shapes the preparation decision.

What do CCDV-F and CCAR-F actually certify?

The Claude Certified Architect, Foundations (CCAR-F) credential certifies systems-level judgment. Its 60 items test whether you can design, orchestrate, and govern Claude-based pipelines at a level of abstraction above individual API calls. You are deciding how agents compose, how tools distribute across roles, how configuration propagates across environments, and how context degradation is prevented at scale.

The Claude Certified Developer, Foundations (CCDV-F) credential certifies application-level execution. Its 53 items test whether you can call Claude correctly, efficiently, and safely from application code. You are constructing API requests, handling responses, managing context windows, evaluating outputs, and integrating Claude into production systems.

Both launched on 12 March 2026 as part of the Claude Partner Network, a $100M programme. As of 3 June 2026, more than 10,000 individuals hold at least one Claude certification across the four tracks. Neither credential is a prerequisite for the other, and the exam guides treat them as parallel entry points to distinct professional tracks.

How do the domain weights compare between the two exams?

Domain weight is the fastest signal of what each exam rewards. The table below maps equivalent skill areas across both credentials. Where a skill exists only on one exam, the other cell is blank.

The contrast is structural. CCAR-F distributes weight across five domains with a 27% ceiling in Agentic Architecture and Orchestration. CCDV-F concentrates weight in a single domain covering 33.1% of the exam, Applications and Integration, before the remaining seven domains share the rest. Mastering CCDV-F Domains 2 and 5 alone gives command of just under 50% of the exam before a candidate opens any other topic.

How do the two exam formats differ?

The scenario bank on CCAR-F is the most operationally significant difference. Each sitting draws 4 scenarios at random from a pool of 6. Two candidates sitting the same day may face the same skill domains but different contextual frames. CCAR-F candidates should work through all six scenario families so no draw catches them unprepared.

CCDV-F has no scenario bank. Per the official CCDV-F exam guide (2026-07-08), items are written directly against the skills in each domain rather than constructed around a pre-built scenario. Item variation comes from the breadth of eight domains rather than from scenario-frame variation across five.

What does the score report tell you after each exam?

Both exams return a pass/fail verdict, a scaled score, and percent-correct by domain. Anthropic does not publish the raw-to-scaled conversion for either exam. On a linear read the CCAR-F pass mark is roughly 41 to 42 correct of 60, but the actual conversion is non-linear. The domain-level percentage breakdown is the most actionable part of the report: it tells you exactly where to invest study time before a retake. Both credentials are valid for 12 months from the award date.

Who should take CCDV-F first, and who should start with CCAR-F?

Take CCDV-F first if you write application code that calls Claude: setting API parameters, handling streaming responses, caching prompts, integrating Claude into data pipelines, or evaluating model outputs. CCDV-F Domain 2 (Applications and Integration, 33.1%) and Domain 5 (Model Selection and Optimisation, 16.8%) directly validate the skills you exercise in production. The developer track is also the stronger fit if your role involves Tool Design and MCP Integration at the implementation level, writing tool definitions and configuring MCP servers from code.

Take CCAR-F first if you design systems that include Claude as a component: specifying agent handoffs, distributing tools across roles, configuring Claude Code for a team, or preventing context degradation at scale. CCAR-F Domain 1 (Agentic Architecture and Orchestration, 27%) and Domain 3 (Claude Code Configuration and Workflows, 20%) reward that systems-design thinking directly.

Some candidates hold both credentials within the same certification cycle. The developer track builds the API fluency that makes architect-level trade-offs concrete; the architect track builds the compositional thinking that makes developer-level decisions more defensible. We see both orderings succeed; the deciding factor is which credential maps most directly to your current role.

Where does the technical depth differ most?

Tool Design and MCP Integration appears on both exams. CCDV-F Domain 8 (10.6%) tests whether you can write tool definitions, configure MCP servers, and handle tool errors in application code. CCAR-F Domain 2 (18%) tests whether you can design the tool distribution strategy across a multi-agent system, including when to split tools, how to scope cross-role access, and how tool descriptions affect model routing behaviour at scale.

Prompt Engineering and Structured Output follows the same pattern. CCDV-F Domain 6 (11.0%) rewards knowing when few-shot examples reduce extraction error and how to manage context windows under token pressure. CCAR-F Domain 4 (20%) applies the same mechanics to compositional prompt chains where a single failure propagates downstream through an agent pipeline.

Context Management and Reliability is a standalone CCAR-F domain at 15%. On CCDV-F it is embedded across other domains. CCAR-F candidates must study context window degradation, summary injection, and session management as a unified body of knowledge.

Claude Code Configuration and Workflows carries 20% on CCAR-F and only 3.1% on CCDV-F. This is the starkest single-domain divergence between the two credentials. Candidates targeting CCAR-F who have not used Claude Code in production should treat this as a substantial investment; those targeting CCDV-F can treat it as low-stakes.

The Agents and Workflows domain on CCDV-F (14.7%) and the Agentic Architecture and Orchestration domain on CCAR-F (27%) cover similar conceptual territory but test different decision layers. CCDV-F items test how you handle tool result errors in a running loop, structure the messages array for multi-turn interactions, or stream responses from a long-running workflow. CCAR-F items test how you design the loop itself: when to add a human checkpoint, how to prevent runaway execution, and how to route errors across a multi-agent system.

What does a practical study split look like for each exam?

Allocating study time in proportion to domain weight is the most mechanically defensible approach. The breakdowns below assume a 12-hour block as the base unit.

For CCAR-F (60 items, 5 domains):

Domain 1  Agentic Architecture and Orchestration    27%    ~3h 14m
Domain 3  Claude Code Configuration                 20%    ~2h 24m
Domain 4  Prompt Engineering and Structured Output  20%    ~2h 24m
Domain 2  Tool Design and MCP Integration           18%    ~2h 09m
Domain 5  Context Management and Reliability        15%    ~1h 48m

For CCDV-F (53 items, 8 domains):

Domain 2  Applications and Integration              33.1%  ~3h 58m
Domain 5  Model Selection and Optimisation          16.8%  ~2h 01m
Domain 1  Agents and Workflows                      14.7%  ~1h 46m
Domain 6  Prompt and Context Engineering            11.0%  ~1h 19m
Domain 8  Tools and MCPs                            10.6%  ~1h 16m
Domain 7  Security and Safety                        8.1%  ~0h 58m
Domain 3  Claude Code                                3.1%  ~0h 22m
Domain 4  Eval, Testing, and Debugging               2.6%  ~0h 19m

On CCDV-F, Domain 2 alone consumes roughly 4 hours per 12-hour block, more than the four lowest-weighted domains combined. That asymmetry is worth internalising before building a study schedule.

Both exams include multiple-response items where the question states explicitly how many options to select. Partial selection scores zero even when some chosen answers are correct. Practise flagging these items and returning rather than guessing under time pressure. CCDV-F candidates have slightly more time per item on average (approximately 136 seconds versus 120 seconds on CCAR-F).

The CCAR-F exam consistently rewards deterministic solutions over probabilistic ones when stakes are high, proportionate fixes, and root-cause tracing, per Anthropic's exam guide. We observe the same preference on CCDV-F evaluation items, where the correct answer traces to the root cause rather than applies a workaround.

Our adaptive study platform covers all 174 atomic concepts mapped to CCAR-F's five domains and 30 task statements at the AI Skill Certs concept library, alongside adaptive practice exams for both CCAR-F and CCDV-F scored 100 to 1000 against the 720 passing bar. AI Skill Certs is an independent prep platform and is not affiliated with or endorsed by Anthropic.