A high-volume classifier sometimes needs deeper judgment. Which design balances cost and quality?
Why D is the answer
Route to a fast model first, and escalate cases that fail validation
Routing gives most traffic a cheap path and keeps the capable model for hard cases. Validation decides which cases escalate.
In detail
- Most traffic takes the cheap path.
- Validation decides which cases escalate to the capable model.
- Track the escalation rate, because a rise is an early warning.
Use case
An illustrative scenario, not a real incident.
A classifier handles millions of messages. A fast model handles each one, and any output failing validation goes to a stronger model, so costs stay low.
See it in different ways
flowchart LR
A["Request"] --> B{"Complexity check"}
B -->|"Simple"| C["Fast model"]
B -->|"Hard"| D["Most capable model"]
C --> E{"Validation passes?"}
E -->|No| D
E -->|Yes| F["Answer"]sequenceDiagram
participant U as Request
participant F as Fast model
participant V as Validation
participant S as Stronger model
U->>F: Handle first
F->>V: Check the output
alt Fails validation
V->>S: Escalate
else Passes
V-->>U: Return the answer
endstateDiagram-v2 state "Fast model" as s0 state "Validation" as s1 state "Done" as s2 state "Stronger model" as s3 [*] --> s0 s0 --> s1 s1 --> s2: passes s1 --> s3: fails s3 --> s2 s2 --> [*]
Key point: Validation decides what escalates.
Why the other options are wrong
- A. Let each user choose a model from a menu before they send a request
- Users cannot be expected to judge each request.
- B. Use the most capable model for all traffic, so that quality is never at risk
- This pays the highest price for cases that do not need it.
- C. Use the cheapest model for everything, and accept the errors as a known cost
- Known errors on hard cases are avoidable with routing.