Game world models have recently demonstrated promising capabilities in generating visually coherent and action-controllable gameplay videos. However, non-player character (NPC) behavior in existing models is either implicitly entangled with video generation or explicitly prescribed through external control signals. Consequently, a game world model has to jointly understand the state, plan the NPC's response and render its visual outcome, limiting its ability to produce responsive and state-aware NPC behavior. The challenge lies in the lack of an explicit interface for state-grounded decision-making. To this end, we introduce WorldMind, to our knowledge the first decoupled framework for state-aware NPC behavior in game world models. WorldMind separates interactive world modeling into four layers: an Understanding Layer that constructs a compact state from generated frames; a Decision Layer that reasons over the compact state to plan the NPC's next action; a Control Layer that translates the actions into temporally aligned conditions; and a Generation Layer that synthesizes their visual outcomes. By reconnecting layers in a closed interaction loop, WorldMind grounds NPC behavior in the evolving game state. We further introduce BOSS-140K, a dataset of gameplay videos paired with rich internal game states, together with an agent that automates the collection at scale. Experiments on BOSS-140K demonstrate reliable compact state reconstruction and mechanics-grounded planning, with WorldMind preferred over the baselines in approximately 70% of pairwise comparisons for its more tactically appropriate and coherent NPC behavior.</p>\n","updatedAt":"2026-08-25T00:55:20.342Z","author":{"_id":"6696347a0818007c1c7a69be","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6696347a0818007c1c7a69be/k-EEb_x9qASVQn3Pr6BES.png","fullname":"White Tea","name":"teawhite","type":"user","isPro":true,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":1,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.9038139581680298},"editors":["teawhite"],"editorAvatarUrls":["https://cdn-avatars.huggingface.co/v1/production/uploads/6696347a0818007c1c7a69be/k-EEb_x9qASVQn3Pr6BES.png"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2608.21439","authors":[{"_id":"6a8ce75f8dd056518b7f53d2","name":"Zhiyang Deng","hidden":false},{"_id":"6a8ce75f8dd056518b7f53d3","name":"Boran Zhang","hidden":false},{"_id":"6a8ce75f8dd056518b7f53d4","name":"Danze Chen","hidden":false},{"_id":"6a8ce75f8dd056518b7f53d5","name":"Yeying Jin","hidden":false}],"mediaUrls":["https://cdn-uploads.huggingface.co/production/uploads/6696347a0818007c1c7a69be/zLxIt7c_rfo0VgSUUHs_1.mp4"],"publishedAt":"2026-08-18T00:00:00.000Z","submittedOnDailyAt":"2026-08-24T00:00:00.000Z","title":"WorldMind: Decoupled Game World Model for State-Aware NPC Behavior","submittedOnDailyBy":{"_id":"6696347a0818007c1c7a69be","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6696347a0818007c1c7a69be/k-EEb_x9qASVQn3Pr6BES.png","isPro":true,"fullname":"White Tea","user":"teawhite","type":"user","name":"teawhite"},"summary":"Game world models have recently demonstrated promising capabilities in generating visually coherent and action-controllable gameplay videos. However, non-player character (NPC) behavior in existing models is either implicitly entangled with video generation or explicitly prescribed through external control signals. Consequently, a game world model has to jointly understand the state, plan the NPC's response and render its visual outcome, limiting its ability to produce responsive and state-aware NPC behavior. The challenge lies in the lack of an explicit interface for state-grounded decision-making. To this end, we introduce WorldMind, to our knowledge the first decoupled framework for state-aware NPC behavior in game world models. WorldMind separates interactive world modeling into four layers: an Understanding Layer that constructs a compact state from generated frames; a Decision Layer that reasons over the compact state to plan the NPC's next action; a Control Layer that translates the actions into temporally aligned conditions; and a Generation Layer that synthesizes their visual outcomes. By reconnecting layers in a closed interaction loop, WorldMind grounds NPC behavior in the evolving game state. We further introduce BOSS-140K, a dataset of gameplay videos paired with rich internal game states, together with an agent that automates the collection at scale. Experiments on BOSS-140K demonstrate reliable compact state reconstruction and mechanics-grounded planning, with WorldMind preferred over the baselines in approximately 70% of pairwise comparisons for its more tactically appropriate and coherent NPC behavior. Project page: https://teawhite.cn/worldmind_projectpage/","upvotes":3,"discussionId":"6a8ce75f8dd056518b7f53d6","projectPage":"https://teawhite.cn/worldmind_projectpage/","githubRepo":"https://github.com/TeaWhiteBro/WorldMind","githubRepoAddedBy":"user","ai_summary":"WorldMind decouples game world modeling into state understanding, decision-making, control, and generation layers to enable explicit, state-aware NPC behavior.","ai_keywords":["game world models","NPC behavior","state-grounded decision-making","closed interaction loop","compact state reconstruction","mechanics-grounded planning"],"ai_summary_model":"thinkingmachines/Inkling-Small","githubStars":0},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"6696347a0818007c1c7a69be","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6696347a0818007c1c7a69be/k-EEb_x9qASVQn3Pr6BES.png","isPro":true,"fullname":"White Tea","user":"teawhite","type":"user"},{"_id":"6a8cead0a5af8ce7f31c62aa","avatarUrl":"/avatars/c152df5cea611542cad1340597ece03a.svg","isPro":false,"fullname":"wuchenfei","user":"yuzhu456358","type":"user"},{"_id":"6a8cf6558fe74663b83ce7bc","avatarUrl":"/avatars/aeaae28210058f9ee731add8c3ed21d3.svg","isPro":false,"fullname":"lili","user":"swoller","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2608/2608.21439.md","query":{}}">
WorldMind: Decoupled Game World Model for State-Aware NPC Behavior
Abstract
WorldMind decouples game world modeling into state understanding, decision-making, control, and generation layers to enable explicit, state-aware NPC behavior.
Game world models have recently demonstrated promising capabilities in generating visually coherent and action-controllable gameplay videos. However, non-player character (NPC) behavior in existing models is either implicitly entangled with video generation or explicitly prescribed through external control signals. Consequently, a game world model has to jointly understand the state, plan the NPC's response and render its visual outcome, limiting its ability to produce responsive and state-aware NPC behavior. The challenge lies in the lack of an explicit interface for state-grounded decision-making. To this end, we introduce WorldMind, to our knowledge the first decoupled framework for state-aware NPC behavior in game world models. WorldMind separates interactive world modeling into four layers: an Understanding Layer that constructs a compact state from generated frames; a Decision Layer that reasons over the compact state to plan the NPC's next action; a Control Layer that translates the actions into temporally aligned conditions; and a Generation Layer that synthesizes their visual outcomes. By reconnecting layers in a closed interaction loop, WorldMind grounds NPC behavior in the evolving game state. We further introduce BOSS-140K, a dataset of gameplay videos paired with rich internal game states, together with an agent that automates the collection at scale. Experiments on BOSS-140K demonstrate reliable compact state reconstruction and mechanics-grounded planning, with WorldMind preferred over the baselines in approximately 70% of pairwise comparisons for its more tactically appropriate and coherent NPC behavior. Project page: https://teawhite.cn/worldmind_projectpage/
Community
Game world models have recently demonstrated promising capabilities in generating visually coherent and action-controllable gameplay videos. However, non-player character (NPC) behavior in existing models is either implicitly entangled with video generation or explicitly prescribed through external control signals. Consequently, a game world model has to jointly understand the state, plan the NPC's response and render its visual outcome, limiting its ability to produce responsive and state-aware NPC behavior. The challenge lies in the lack of an explicit interface for state-grounded decision-making. To this end, we introduce WorldMind, to our knowledge the first decoupled framework for state-aware NPC behavior in game world models. WorldMind separates interactive world modeling into four layers: an Understanding Layer that constructs a compact state from generated frames; a Decision Layer that reasons over the compact state to plan the NPC's next action; a Control Layer that translates the actions into temporally aligned conditions; and a Generation Layer that synthesizes their visual outcomes. By reconnecting layers in a closed interaction loop, WorldMind grounds NPC behavior in the evolving game state. We further introduce BOSS-140K, a dataset of gameplay videos paired with rich internal game states, together with an agent that automates the collection at scale. Experiments on BOSS-140K demonstrate reliable compact state reconstruction and mechanics-grounded planning, with WorldMind preferred over the baselines in approximately 70% of pairwise comparisons for its more tactically appropriate and coherent NPC behavior.
Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images
Cite arxiv.org/abs/2608.21439 in a model README.md to link it from this page.
Cite arxiv.org/abs/2608.21439 in a dataset README.md to link it from this page.
Cite arxiv.org/abs/2608.21439 in a Space README.md to link it from this page.
Discussion (0)
Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.
Sign in →No comments yet. Sign in and be the first to say something.