Anthropic CCDV-F (Claude Certified Developer - Foundations) Practice Test 27

Question 260

A large-scale tagger processes easy, short classifications and is dominated by latency and cost. Which model-selection tendency is sensible?

  1. Contrast feature parity, quotas, authentication, and region availability for that route.
  2. Assess a Sonnet-class model as a quality-latency compromise.
  3. Execute workload-specific evaluations measuring quality, latency, and total cost.
  4. Commit to the absolute lowest-latency model without testing classification quality.
  5. Benchmark a fast Haiku-class model against the required accuracy threshold.

Explanation

Correct Answer

E:

Lightweight tasks may not need higher-cost reasoning; measured acceptance quality should decide whether the speed benefit is acceptable.

Incorrect Answers

A:

Equivalent model names do not imply equivalent API surfaces or governance controls on every hosting provider; where inexpensive simple classification matters, a control for provider feature parity leaves that test unmet.

B:

The middle tier is designed for balanced workloads; task-specific tests should confirm that it meets the acceptance targets; testing inexpensive simple classification would still fail after improving balanced assistant model baseline.

C:

Public benchmarks may not represent proprietary inputs or tools; a measured tradeoff is more defensible than a brand label; because production model bakeoff is its concern, inexpensive simple classification remains outside its effect here.

D:

Speed alone does not establish that the tagger meets precision and recall requirements. A smaller model should be benchmarked against acceptance thresholds before being selected.

Question 261

An agent must handle difficult multistep reasoning with costly mistakes. What evaluation candidate deserves attention?

  1. Test a capable Opus-class model for quality under the actual workflow.
  2. Verify feature compatibility and run version-specific request tests before rollout.
  3. Verify that the chosen model and provider route support the required input modality.
  4. As an alternative, benchmark a fast Haiku-class model against the required accuracy threshold.

Explanation

Correct Answer

A:

High-capability models can be valuable where reasoning reliability dominates, but production choice still requires measured cost and latency.

Incorrect Answers

B:

Request fields that worked on a prior generation may fail or behave differently; model migration includes API contract validation; when the goal is complex reasoning model selection, optimizing reasoning config compatibility misses that operational boundary.

C:

A low-cost text model is unsuitable if the deployed endpoint cannot process images as required; the consequence is improved multimodal capability match, not verified complex reasoning model selection.

D:

Lightweight tasks may not need higher-cost reasoning; measured acceptance quality should decide whether the speed benefit is acceptable; it improves inexpensive simple classification only, leaving complex reasoning model selection unproven.

Question 262

A conversational assistant balances broad coding quality with response speed and budget. What model family is a reasonable first baseline?

  1. Verify supported context capacity and test performance on representative long inputs.
  2. Evaluate a Sonnet-class model as a quality-latency compromise.
  3. Perform workload-specific evaluations measuring quality, latency, and total cost.
  4. For this workflow, benchmark a fast Haiku-class model against the required accuracy threshold.

Explanation

Correct Answer

B:

The middle tier is designed for balanced workloads; task-specific tests should confirm that it meets the acceptance targets.

Incorrect Answers

A:

The advertised window is a hard capacity consideration, but empirical retrieval and reasoning quality on long cases also matter; this targets long-context model fit, while the system must demonstrate balanced assistant model baseline.

C:

Public benchmarks may not represent proprietary inputs or tools; a measured tradeoff is more defensible than a brand label; where balanced assistant model baseline matters, a control for production model bakeoff leaves that test unmet.

D:

Lightweight tasks may not need higher-cost reasoning; measured acceptance quality should decide whether the speed benefit is acceptable; changing inexpensive simple classification does not establish balanced assistant model baseline in this application.

Question 263

A team chooses a model solely from marketing benchmarks. What should the decision process add?

  1. Run workload-specific evaluations measuring quality, latency, and total cost.
  2. Review a Sonnet-class model as a quality-latency compromise.
  3. Check supported context capacity and test performance on representative long inputs.
  4. Evaluate a capable Opus-class model for quality under the actual workflow.

Explanation

Correct Answer

A:

Public benchmarks may not represent proprietary inputs or tools; a measured tradeoff is more defensible than a brand label.

Incorrect Answers

B:

The middle tier is designed for balanced workloads; task-specific tests should confirm that it meets the acceptance targets; it changes balanced assistant model baseline; it does not prove production model bakeoff in this case.

C:

The advertised window is a hard capacity consideration, but empirical retrieval and reasoning quality on long cases also matter; a remedy for long-context model fit is not a substitute for production model bakeoff.

D:

High-capability models can be valuable where reasoning reliability dominates, but production choice still requires measured cost and latency; for production model bakeoff, this instead manages complex reasoning model selection; the requested decision remains unresolved.

Question 264

A product supports screenshots in prompts. What must model selection verify besides nominal intelligence?

  1. Check that the chosen model and provider route support the required input modality.
  2. As an alternative, route qualifying easy workloads to the lower-cost model and monitor quality.
  3. At the application layer, benchmark a fast Haiku-class model against the required accuracy threshold.
  4. Inspect feature compatibility and run version-specific request tests before rollout.

Explanation

Correct Answer

A:

A low-cost text model is unsuitable if the deployed endpoint cannot process images as required. Verify modality support for the exact deployed model and provider route, including any per-image limits and supported formats.

Incorrect Answers

B:

Measured equivalence allows savings; indiscriminate downgrading of high-risk cases would be unsound; testing multimodal capability match would still fail after improving multi-criterion integration.

C:

Lightweight tasks may not need higher-cost reasoning; measured acceptance quality should decide whether the speed benefit is acceptable; this targets inexpensive simple classification, while the system must demonstrate multimodal capability match.

D:

Request fields that worked on a prior generation may fail or behave differently; model migration includes API contract validation; the consequence is improved reasoning config compatibility, not verified multimodal capability match.

Question 265

A new model release looks stronger but changes supported thinking parameters. How should migration proceed?

  1. Inspect that the chosen model and provider route support the required input modality.
  2. Test a Sonnet-class model as a quality-latency compromise.
  3. Employ a fast default model with measured escalation to a stronger model when justified.
  4. Check feature compatibility and run version-specific request tests before rollout.

Explanation

Correct Answer

D:

Request fields that worked on a prior generation may fail or behave differently; model migration includes API contract validation.

Incorrect Answers

A:

A low-cost text model is unsuitable if the deployed endpoint cannot process images as required; where reasoning config compatibility matters, a control for multimodal capability match leaves that test unmet.

B:

The middle tier is designed for balanced workloads; task-specific tests should confirm that it meets the acceptance targets; this may help balanced assistant model baseline but cannot by itself deliver reasoning config compatibility.

C:

Tiered routing spends expensive inference only on tasks that need it, provided the escalation classifier is tested; this targets adaptive model routing, while the system must demonstrate reasoning config compatibility.

Question 266

A support application occasionally handles extremely long case files. What should guide model choice?

  1. For this workflow, route qualifying easy workloads to the lower-cost model and monitor quality.
  2. Exercise a capable Opus-class model for quality under the actual workflow.
  3. Examine that the chosen model and provider route support the required input modality.
  4. Analyze a Sonnet-class model as a quality-latency compromise.
  5. Confirm supported context capacity and test performance on representative long inputs.

Explanation

Correct Answer

E:

The advertised window is a hard capacity consideration, but empirical retrieval and reasoning quality on long cases also matter.

Incorrect Answers

A:

Measured equivalence allows savings; indiscriminate downgrading of high-risk cases would be unsound; it follows multi-criterion integration constraints but leaves long-context model fit unspecified.

B:

High-capability models can be valuable where reasoning reliability dominates, but production choice still requires measured cost and latency; this implements complex reasoning model selection rather than the demanded long-context model fit.

C:

A low-cost text model is unsuitable if the deployed endpoint cannot process images as required; testing long-context model fit would still fail after improving multimodal capability match.

D:

The middle tier is designed for balanced workloads; task-specific tests should confirm that it meets the acceptance targets; for long-context model fit, this instead manages balanced assistant model baseline; the requested decision remains unresolved.

Question 267

A chatbot must stay within a low median latency SLA but can escalate rare difficult queries. Which routing design is appropriate?

  1. At the application layer, route qualifying easy workloads to the lower-cost model and monitor quality.
  2. Validate a capable Opus-class model for quality under the actual workflow.
  3. Use a fast default model with measured escalation to a stronger model when justified.
  4. Evaluate feature parity, quotas, authentication, and region availability for that route.

Explanation

Correct Answer

C:

Tiered routing spends expensive inference only on tasks that need it, provided the escalation classifier is tested.

Incorrect Answers

A:

Measured equivalence allows savings; indiscriminate downgrading of high-risk cases would be unsound; it changes multi-criterion integration; it does not prove adaptive model routing in this case.

B:

High-capability models can be valuable where reasoning reliability dominates, but production choice still requires measured cost and latency; this targets complex reasoning model selection, while the system must demonstrate adaptive model routing.

D:

Equivalent model names do not imply equivalent API surfaces or governance controls on every hosting provider; for adaptive model routing, this instead manages provider feature parity; the requested decision remains unresolved.

Question 268

A team locks itself to one model ID without a replacement plan. Which operational risk remains?

  1. A pinned model ID guarantees indefinite availability without vendor migration work.
  2. Switching to any new model name preserves identical tool and output behavior.
  3. A prompt cache can preserve an unavailable inference model after retirement.
  4. Setting a constant temperature eliminates future model-version dependencies.
  5. A version may be deprecated or retired, requiring tested migration and fallback.
  6. A missing migration plan matters only when training the foundation model.

Explanation

Correct Answer

E:

Pinning improves reproducibility but does not prevent lifecycle retirement; compatibility planning is still needed.

Incorrect Answers

A:

Versioned deployment identifiers may still be retired or restricted according to their lifecycle. A retirement schedule can force a production migration.

B:

Feature and parameter compatibility must be tested because family and version differences can affect requests.

C:

Caching reusable input prefixes does not operate as a model hosting or lifecycle support mechanism. A working model deployment must still exist to process requests.

D:

A decoding parameter cannot preserve API availability or make schemas and behavior identical across upgrades. A replacement needs its own capability and behavior regression suite.

F:

Application teams must test and deploy replacement models even when they never train model weights. The team remains responsible for release compatibility and continuity.

Question 269

A provider route exposes Claude through a managed cloud platform. What should the architect verify before switching from the native API?

  1. Validate that the chosen model and provider route support the required input modality.
  2. Compare feature parity, quotas, authentication, and region availability for that route.
  3. In this design, benchmark a fast Haiku-class model against the required accuracy threshold.
  4. Adopt a fast default model with measured escalation to a stronger model when justified.

Explanation

Correct Answer

B:

Equivalent model names do not imply equivalent API surfaces or governance controls on every hosting provider. Compare authentication, region availability, request syntax, model identifiers, tool support, and observability under the target provider.

Incorrect Answers

A:

A low-cost text model is unsuitable if the deployed endpoint cannot process images as required; it changes multimodal capability match; it does not prove provider feature parity in this case.

C:

Lightweight tasks may not need higher-cost reasoning; measured acceptance quality should decide whether the speed benefit is acceptable; the consequence is improved inexpensive simple classification, not verified provider feature parity.

D:

Tiered routing spends expensive inference only on tasks that need it, provided the escalation classifier is tested; for this request, provider feature parity is the missing piece, not adaptive model routing.

Question 270

A high-volume service repeats long instructions and runs latency-insensitive evaluations overnight. Which TWO optimizations address those distinct cost drivers? Choose TWO.

  1. Store full debug logs in every subsequent prompt.
  2. Cache the stable repeated instruction prefix.
  3. Send independent offline evaluations through Message Batches.
  4. Use synchronous user-facing streaming for every offline job.
  5. Insert a changing timestamp at the beginning of every cached prompt.

Explanation

Correct Answers

B:

Prompt caching reduces repeated input processing when the prefix matches and qualifies for caching. Put invariant instructions ahead of variable user data so cache reuse is preserved across eligible requests.

C:

Asynchronous batch pricing is favorable when immediate results are not required. Queue independent tasks asynchronously and reconcile results with stable identifiers after batch completion.

Incorrect Answers

A:

Unnecessary transcript replay increases input-token consumption. Use telemetry to select relevant excerpts rather than feeding every historical debugging record back to the model.

D:

Streaming improves presentation latency but does not exploit deferred batch processing. Client-side streaming does not turn an interactive request into a discounted asynchronous batch job.

E:

Prefix variation destroys cache reuse instead of improving it. Moving changing material after the reusable prefix is more compatible with prefix-based cache matching.

Claude Certified Developer - Foundations (CCDV-F) Practice Tests