# Benchmark: Kitchen Appliances Selection

**Topic:** Я хочу найти и выбрать между встраиваемыми вытяжкой + варочной панелью индукционной + духовым электрическим шкафом модели, у которых есть физические элементы управления, хорошие отзывы, высокая надежность и обслуживание с доступными гарантийными запчастями в РФ в среднем ценовом сегменте.

**Date:** 2026-05-12

---

## Experiments

| # | Method | Cost | Time | Quality Score |
|---|--------|------|------|---------------|
| 1 | Perplexity Deep Research | ? | ? | ? |
| 2 | Extended Search (old) | ? | ? | ? |
| 3 | Extended Search (optimized) | ? | ? | ? |

---

## Metrics Definition

### Cost
- Total $ charged for LLM calls
- Include: INPUT, OUTPUT, CACHE tokens

### Time
- Wall time from start to final report (minutes)

### Quality Score
- **Coverage:** Did we find options for all 3 appliance types? (0-3 points)
- **Depth:** Are there specific model recommendations with pros/cons? (0-3 points)
- **Criteria Match:** Physical controls, reliability, serviceability in RF? (0-3 points)
- **Actionability:** Can user make a decision? (0-1 point)

**Max score: 10**

---

## Files Structure

```
benchmarks/kitchen-appliances/
├── README.md (this file)
├── exp-01-perplexity/
│   ├── report.md
│   └── metrics.json
├── exp-02-extended-old/
│   ├── report.md
│   ├── metrics.json
│   └── steps/
├── exp-03-extended-optimized/
│   ├── report.md
│   ├── metrics.json
│   └── steps/
└── comparison.md
```