PEC-Home (RAG) - Dynamic Preferences Level 1: leaderboard
Metric: Execution accuracy (%; 712 smart-home commands at ellipsis level 1 of 4 (complete commands), functionally correct device operations; dialogue history retrieved from a vector store, top-3). Source: arxiv.org. Saturation forecast: Estimated already saturated. 10 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 2.5 7B Instruct | 93.68 |
| 2 | Gemma 2 27B (IT) | 92.98 |
| 3 | Gemma 2 9B (IT) | 92.13 |
| 4 | GPT-4o | 91.57 |
| 5 | Qwen 2.5 32B Instruct | 90.31 |
| 6 | Qwen 2.5 72B Instruct | 84.55 |
| 7 | DeepSeek V3 | 80.62 |
| 8 | Mistral 7B Instruct (v0.3) | 79.92 |
| 9 | Qwen 2.5 14B Instruct | 79.21 |
| 10 | Llama 3.1 8B Instruct | 74.3 |
Interactive version: theaggregate.ai/benchmark?slug=pec-home-rag-dynamic-preferences-level-1 · How It Works · Data refreshed daily, snapshot 2026-09-26.