--- format: "aidr-story-markdown/v1" id: "71f077148c91c99f7723ac6854539d22f1b1502d98de8e2fc1da6cadfa4109bb" canonical_url: "https://aidr.today/71f07714?lang=en" title: "AMI Launches First Hierarchical World Model for Long Horizon Visual Planning" lang: "en" requested_lang: "en" available_langs: ["en","vi"] translation_fallback: null fallback_fields: [] published_at: "2026-10-06T16:08:40.000Z" category: "Models" topics: ["agent"] source_urls: ["https://huggingnews.com/ai/ami-launches-first-hierarchical-world-model-for-long-horizon-visual-plan-8645b2a0","https://marketbrief.now/ai/ami-launches-first-hierarchical-world-model-for-long-horizon-visual-plan-8645b2a0"] summary: "AMI launched H-JEPA, which it describes as the first end-to-end learned hierarchical world model for long horizon visual planning. The model improved success on the Visual Antmaze benchmark from 18% to 73% while reducing planning compute, learning world models at various abstraction levels and planning from coarse goals to finer actions. The architecture is built atop SIGReg/LeWM and uses a stable training recipe that allows semantic abstractions to emerge naturally across its levels." --- # AMI Launches First Hierarchical World Model for Long Horizon Visual Planning > [Open the canonical story]() **Published:** 2026-10-06T16:08:40.000Z **Category:** Models **Topics:** agent ## Summary AMI launched H\-JEPA, which it describes as the first end\-to\-end learned hierarchical world model for long horizon visual planning\. The model improved success on the Visual Antmaze benchmark from 18% to 73% while reducing planning compute, learning world models at various abstraction levels and planning from coarse goals to finer actions\. The architecture is built atop SIGReg/LeWM and uses a stable training recipe that allows semantic abstractions to emerge naturally across its levels\. ## Sources - [Story source]() - [Story source]()