PersonalHomeBench - Information Grounding (Agentic): leaderboard

PersonalHomeBench (2026) tests models as assistants in generated, persona-rich smart homes. Information grounding asks factual and multi-hop questions about the household; accuracy in percent, acting through the benchmark household toolbox to gather information. Higher is better.

Metric: Accuracy (%) on 1,000 information grounding questions (factual lookup to multi-hop reasoning over household entities; exact match, else a Gemini 2.5 Pro semantic-equivalence judge), over the text-based split of PersonalHomeBench (1,100 generated, manually reviewed households with persona-based occupants and appliances), temperature 0, in the agentic setting (the model must gather household information and act through the benchmark's PersonalHomeTools toolbox, up to 15 turns); each model at its stated reasoning mode; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 7 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1GPT-473.8#970
2GPT-4o71.03#751
3GPT-OSS-20BHigh69.6#1183
4Qwen 3 30BThinking67.5
5nemotron-3-nano-30B-a3B64.9#1946
6Qwen 3 30B Instruct64.14#1365
7Qwen 3 4BNon-reasoning56.4#2348

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=personalhomebench-information-grounding-agentic · How It Works · Data refreshed daily, snapshot 2026-11-21.