PEC-Home (ICL) - Multi-User Level 1: leaderboard
Smart-home command interpretation in PEC-Home (Shan et al., 2026), multi-user scenario, where household members expect different settings: complete commands (ellipsis level 1 of 4), with the model prompted with one in-context example per ellipsis level but without the dialogue history. Execution accuracy in percent, the share of commands mapped to the intended device operations; higher is better.
Metric: Execution accuracy (%; 1,068 smart-home commands at ellipsis level 1 of 4 (complete commands), functionally correct device operations; one in-context example per ellipsis level, no dialogue history). Source: arxiv.org. Saturation forecast: Estimated already saturated. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemma 2 27B (IT) | 93.63 | #1815 |
| 2 | Gemma 2 9B (IT) | 93.26 | #2275 |
| 3 | GPT-4o | 92.98 | #741 |
| 4 | DeepSeek V3 | 92.79 | #705 |
| 5 | Qwen 2.5 32B Instruct | 92.13 | #1216 |
| 6 | Mistral 7B Instruct (v0.3) | 90.36 | #6698 |
| 7 | Qwen 2.5 72B Instruct | 89.42 | #1093 |
| 8 | Qwen 2.5 7B Instruct | 88.58 | #2583 |
| 9 | Qwen 2.5 14B Instruct | 85.86 | #1745 |
| 10 | Llama 3.1 8B Instruct | 79.21 | #4277 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=pec-home-icl-multi-user-level-1 · How It Works · Data refreshed daily, snapshot 2026-11-12.