{"slug": "openhome-from-perception-to-insight-in-smart-home-camera-videos", "title": "OpenHome: From Perception to Insight in Smart-Home Camera Videos", "summary": "A new benchmark, OpenHome, shows frontier closed-source vision language models outperform open-weight models on every capability level for smart-home camera video understanding, with the best model, Gemini 4 Argon, reaching 81.2% overall accuracy. OpenHome contains 2,254 four-option multiple-choice questions across 1,118 real-world video clips, organized into 10 tasks spanning Perception, Reasoning and Insight. The best closed-source model leads the best open-weight model by 16.3% on Perception, 22.4% on Reasoning and 37.4% on Insight, according to results reported for 32 closed-source and open-weight VLMs.", "body_md": "arXiv:2610.31986v1 Announce Type: new \nAbstract: Vision language models (VLMs) have been increasingly adopted in consumer smart-home camera systems to interpret recorded videos and provide insights. However, video clips captured from smart-home cameras are different from curated web videos dominating existing VLM benchmarks. Smart-home video clips often do not contain subtitles or narration and are captured from different viewing angles, under various lighting conditions, and in diverse scenarios. However, the capabilities of the model in this domain, under sparse visual signals and limited context, are not fully assessed by existing VLM benchmarks. To close this gap, we introduce OpenHome, a visual-only, four-option multiple-choice benchmark for evaluating smart-home video understanding. It contains 2,254 questions on 1,118 real-world video clips and organizes 10 tasks into three capability levels. Perception evaluates visible content, Reasoning evaluates causal, spatial, and temporal relationships, and Insight evaluates subject intent, supporting evidence and appropriate homeowner responses. Based on the dataset, we describe the task design, data pipeline and evaluation protocol, and report results for 32 closed-source and open-weight VLMs. The best model, Gemini 4 Argon, reaches 81.2% overall accuracy. Frontier closed-source models outperform open-weight models on every level. The best closed-source model leads the best open-weight model by 16.3% on Perception, 22.4% on Reasoning and 37.4% on Insight.", "url": "https://wpnews.pro/news/openhome-from-perception-to-insight-in-smart-home-camera-videos", "canonical_source": "https://arxiv.org/abs/2610.31986", "published_at": "2026-10-26 04:00:00+00:00", "updated_at": "2026-10-26 04:12:29.502489+00:00", "lang": "en", "topics": ["computer-vision", "ai-research", "large-language-models", "ai-products"], "entities": ["OpenHome", "Gemini 4 Argon", "arXiv"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/openhome-from-perception-to-insight-in-smart-home-camera-videos", "markdown": "https://wpnews.pro/news/openhome-from-perception-to-insight-in-smart-home-camera-videos.md", "text": "https://wpnews.pro/news/openhome-from-perception-to-insight-in-smart-home-camera-videos.txt", "jsonld": "https://wpnews.pro/news/openhome-from-perception-to-insight-in-smart-home-camera-videos.jsonld"}}