Multi-turn long-horizon planning is critical for foundation model agents, yet how to fundamentally improve it remains unclear. Existing models are trained on uncontrollable and opaque Internet data, making it difficult to identify how planning ability is acquired, shaped, and integrated.
arXiv:2607. 24720v1 Announce Type: cross Abstract: Multi-turn long-horizon planning is critical for foundation model agents, yet how to fundamentally improve it remains unclear.
The paper introduces the Dual-Latent World Model (Dual-WM), which separates local execution and long-range planning into distinct latent spaces and dynamics models. A new learning method, Long-Horizon Representation Learning with Weighted Rollout (LoRe), supervises predictions at both levels using exponential horizon weights. Experiments on five goal-conditioned visual control tasks show that Dual-WM improves success rates over strong baselines, especially at longer horizons.
arXiv:2608. 11967v1 Announce Type: cross Abstract: Large language model agents increasingly rely on long-horizon reasoning to solve complex tasks involving planning, tool use, and memory.
arXiv:2609.38334v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly deployed as agents for multi-step decision-making, yet transfer poorly to unseen environments. World-mode...
arXiv:2606. 27483v1 Announce Type: new Abstract: Large language model (LLM) agents have demonstrated strong capability in sequential decision-making, yet they remains fundamentally reactive in long-horizon tasks.
Reinforced Planning with Latent World Models (RP1) is a novel method that learns to evaluate imagined outcomes via a critic and to improve multi‑step plans through an optimizer trained offline on world‑model roll‑outs. It is the first approach to fully learn plan improvement and can be attached to any pretrained latent world model. In experiments on visual navigation, arm reaching, and robotic manipulation, RP1 outperforms hand‑designed search algorithms, achieving near‑perfect success while using far fewer roll‑outs and running up to 67× faster than the strongest alternative.
arXiv:2605. 08732v2 Announce Type: replace-cross Abstract: Modern vision-based world models can represent observations as compact yet expressive latent manifolds, but fast goal-oriented planning in these spaces remains challenging.
The paper introduces temporal self‑distillation for reinforcement learning with verifiable rewards (RLVR), proposing that a policy can learn from a stronger future checkpoint of itself. Two methods—Near‑Future Policy Optimization (NPO) and Near‑Future Policy Distillation (NPD)—use verified future‑self trajectories and token‑level transfer, respectively, while AutoNPO adaptively selects the optimal future checkpoint. Experiments on eight image‑text benchmarks show that near‑future teachers yield higher performance than far‑future ones, indicating that the balance between new capability and learner compatibility is key.
arXiv:2606. 27814v4 Announce Type: replace Abstract: Training small language-model agents for long-horizon interactive tasks requires both fast imitation and reward-driven improvement.
Reinforced Planning with Latent World Models introduces RP1, a neural planner that learns to evaluate imagined outcomes via a critic and improve multi‑step plans through an optimizer trained offline on world‑model roll‑outs. Unlike existing planners that are hand‑designed or only inform policies, RP1 fully learns to refine plans and can be attached to any pretrained latent world model. In experiments on visual navigation, arm reaching, and robotic manipulation, RP1 outperforms hand‑designed search algorithms, achieving near‑perfect success while using 1,000× fewer roll‑outs and up to 67× faster inference.
arXiv:2606. 27814v1 Announce Type: new Abstract: Training small language-model agents for long-horizon interactive tasks requires both fast imitation and reward-driven improvement.
arXiv:2606. 15912v1 Announce Type: cross Abstract: Multi-turn agents that plan, invoke tools, and interact with environments offer a promising paradigm for solving complex tasks, yet their capabilities typically rely on very large models whose inference cost is prohibitive in practice.