PEC-Home (RAG) - Dynamic Preferences Level 4: leaderboard
Metric: Execution accuracy (%; 712 smart-home commands at ellipsis level 4 of 4 (progressively elliptical), 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 | GPT-4o | 50.7 |
| 2 | DeepSeek V3 | 49.58 |
| 3 | Qwen 2.5 72B Instruct | 45.65 |
| 4 | Qwen 2.5 32B Instruct | 43.68 |
| 5 | Qwen 2.5 7B Instruct | 43.4 |
| 6 | Gemma 2 27B (IT) | 43.12 |
| 7 | Qwen 2.5 14B Instruct | 40.73 |
| 8 | Mistral 7B Instruct (v0.3) | 33.99 |
| 9 | Gemma 2 9B (IT) | 32.72 |
| 10 | Llama 3.1 8B Instruct | 24.02 |
Interactive version: theaggregate.ai/benchmark?slug=pec-home-rag-dynamic-preferences-level-4 · How It Works · Data refreshed daily, snapshot 2026-09-26.