PEC-Home (RAG) - Multi-User Level 4: leaderboard
Metric: Execution accuracy (%; 1,068 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 | 55.71 |
| 2 | DeepSeek V3 | 53.37 |
| 3 | Qwen 2.5 72B Instruct | 52.25 |
| 4 | Qwen 2.5 32B Instruct | 48.97 |
| 5 | Qwen 2.5 7B Instruct | 48.5 |
| 6 | Qwen 2.5 14B Instruct | 48.41 |
| 7 | Mistral 7B Instruct (v0.3) | 47.47 |
| 8 | Gemma 2 27B (IT) | 45.04 |
| 9 | Llama 3.1 8B Instruct | 42.88 |
| 10 | Gemma 2 9B (IT) | 38.3 |
Interactive version: theaggregate.ai/benchmark?slug=pec-home-rag-multi-user-level-4 · How It Works · Data refreshed daily, snapshot 2026-09-26.