{"slug": "4dmemory-benchmarking-long-horizon-spatiotemporal-memory-with-harness-augmented", "title": "4DMemory: Benchmarking Long-Horizon Spatiotemporal Memory with Harness-Augmented Agents", "summary": "A new arXiv paper (2610.32332v1) introduces 4DMemoryBench and the LongVideo dataset, comprising about 900 hours of video and over 5,000 manually reviewed questions testing long-horizon spatiotemporal reasoning, and reports that 13 leading models score more than 10% lower on 4DMemoryBench than on a previous benchmark. The authors also present the 4DMemory Harness, which combines four complementary skills, 22+ specialized tool models and a model-specific retrieval-augmented generation system, and report it improves every one of the 13 evaluated models by over 50%. The work targets persistent, fine-grained memory for spatiotemporal agents handling tasks spanning hours to years.", "body_md": "arXiv:2610.32332v1 Announce Type: cross \nAbstract: Humans efficiently process multimodal streams spanning hours to years, building persistent, fine-grained spatiotemporal memories that support reasoning about and predicting long-horizon events. In contrast, existing benchmarks focus on short- and medium-context tasks, and most foundation models struggle to preserve fine-grained information over long horizons. To evaluate and improve long-horizon spatiotemporal reasoning, we introduce the LongVideo dataset and the 4DMemoryBench. LongVideo comprises about 900 hours of video and over 5,000 manually reviewed questions covering visual, textual, and spatiotemporal reasoning across diverse context lengths. We further develop the 4DMemory Harness, which integrates four complementary skills, 22+ specialized tool models, and a model-specific retrieval-augmented generation system. Evaluating 13 leading models, we find that standalone foundation models struggle with long-horizon tasks, while the 4DMemory Harness improves all of them by over 50\\%. Compared to previous benchmark, the same models exhibit a performance drop of more than 10\\% on 4DMemoryBench. We hope 4DMemory will facilitate the development of spatiotemporal agents capable of addressing complex, long-horizon tasks.", "url": "https://wpnews.pro/news/4dmemory-benchmarking-long-horizon-spatiotemporal-memory-with-harness-augmented", "canonical_source": "https://www.machinebrief.com/news/4dmemory-benchmarking-long-horizon-spatiotemporal-memory-wit-evuj", "published_at": "2026-10-26 04:00:00+00:00", "updated_at": "2026-10-26 07:42:28.950802+00:00", "lang": "en", "topics": ["artificial-intelligence", "machine-learning", "ai-research", "ai-agents", "computer-vision"], "entities": ["4DMemoryBench", "LongVideo", "4DMemory Harness", "arXiv"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/4dmemory-benchmarking-long-horizon-spatiotemporal-memory-with-harness-augmented", "markdown": "https://wpnews.pro/news/4dmemory-benchmarking-long-horizon-spatiotemporal-memory-with-harness-augmented.md", "text": "https://wpnews.pro/news/4dmemory-benchmarking-long-horizon-spatiotemporal-memory-with-harness-augmented.txt", "jsonld": "https://wpnews.pro/news/4dmemory-benchmarking-long-horizon-spatiotemporal-memory-with-harness-augmented.jsonld"}}