Long-Horizon Tasks

Large language models often reason at length before answering, increasing cost and latency. Prompts and trained settings can shorten this reasoning, but a shorter trace may only show that the model stopped sooner. Here, we evaluate paired runs of the same question at matched reasoning horizons across 198 GPQA Diamond and 500 MMLU-Pro questions. We test a numeric/concision prompt that announces a token limit for Qwen3-14B and the trained effort settings of gpt-oss-20b and -120b. The Qwen prompt shortens reasoning traces by 12-17%, while accuracy changes at matched token limits are small and mixed. A concise/early-answer instruction raises MMLU-Pro accuracy by 3.8 percentage points at 512 tokens, including +2.7 points when both runs are unfinished. Its gain at 2,048 tokens is uncertain. For gpt-oss, candidate-logit answers from completed low- and medium-effort reasoning are 14.5-26.3 points more accurate than matched-horizon high-effort answers. Most of the 512-token advantage comes from lower effort finishing earlier, while differences among unfinished runs are smaller and mixed. Wrong early answers often concentrate probability on the chosen option, so earlier stopping does not uniformly improve probability quality. In these tests, a tight deadline can favor lower effort or a concise instruction, whereas allowing high effort to finish can recover higher final accuracy. Evaluations should report correct completion before a deadline, the answer obtained when a run is stopped, differences among unfinished runs, and probability assigned to the correct answer separately.

Humans perform long-horizon manipulation by retaining knowledge of what earlier actions have established while continuously adapting the motion underway. By contrast, action-chunked vision-language-action (VLA) policies repeatedly replan from the current input at each query. Existing methods preserve either long-term task evidence through memory or short-term motion through action reuse and ensembling, leaving the cross-query handoff incomplete. We introduce ChainVLA, a 1.2B-parameter VLA policy that chains successive queries through a joint and revisable execution state. Progress Context combines a recurrent Working State with sparse event memory to carry observation-derived task progress, while Motion Tail feeds the preceding prediction's unexecuted continuation into state construction and action generation. Together, the two components condition a decoder that regenerates each action horizon under the latest observation, allowing the carried state to guide the next prediction without fixing it. ChainVLA reaches 62.8% average success on RMBench and 98.8% across four LIBERO suites, while removing Motion Tail or Progress Context reduces RMBench success to 11.2% and 3.0%, respectively. These asymmetric ablations are consistent with motion continuity helping preserve the observation stream from which task progress is inferred.

Large language model (LLM) agents increasingly undertake long-horizon tasks that require sustained reasoning, tool use, and revision across many interdependent steps. However, existing agent harnesses maintain task execution, task state, and completion assessment within a growing context, making the state difficult to track and allowing incorrect self-assessments to propagate into later decisions. We reformulate long-horizon execution as a task-state management problem and propose LongHorizon-Harness, which maintains the task state explicitly outside execution and updates it only with facts independently verified from the environment. Its Manage-Execute-Audit(MEA) loop uses a manager to maintain the task state and determine the next subtask, a fresh-context executor to perform it, and a read-only auditor to verify the resulting environment state before the next round. A lightweight AgentAdapter supports interchangeable model and harness backends without modifying their native agent loops. LongHorizon-Harness improves Qwen3.7-Plus from 51.8% to 80.7% on WeaveBench, from 69.7% to 77.2% on Terminal-Bench2.1, and from 2.8% to 8.3% on OSWorld2.0. It also raises Claude Opus4.7 from 20.0% to 34.3% on an OSWorld2.0 subset, demonstrating consistent gains across models, harnesses, and interaction domains.

The success of coding agents has established the harness as a paradigm: what an agent achieves depends not on the model alone, but on the infrastructure around it. We ask whether the same paradigm extends to embodied agents in the physical world. We present Thea, a harness in which an agentic loop orchestrates robot capabilities, each wrapped as a callable tool. It inherits the core components of coding agents, modified as the physical world requires. The world, however, withholds two abilities that software grants for free: reading the state of the world, and judging the outcome of an action. To bridge these gaps, Thea introduces Scene Graph as Context, a persistent, symbolic representation of the world, and Evaluation as Exit Codes, which detects when an action should terminate, judges whether it succeeded, and on failure diagnoses the cause. Together they close the loop between the agent and the physical world. Rich behaviors then emerge from the composition of tools, and the closed loop carries long-horizon tasks to completion in real environments.

JEPA-style visual world models offer an effective paradigm for visual goal planning by predicting future latent representations. Existing methods typically learn local transition consistency through next-step representation prediction. However, in long-horizon tasks, accurate local prediction alone need not ensure sustained progress toward the goal. First, multi-step rollouts can remain locally plausible while drifting away from goal-relevant trajectories. Second, locally similar future states can correspond to substantially different long-term progress, making them difficult to distinguish in a latent space optimized mainly for local consistency. To address these challenges, we introduce goal-conditioned progress order, a relative ordering of states according to how they advance toward a given goal. This order exhibits an asymmetric, coarse-to-fine structure: early states retain broader future possibilities, while later states concentrate on more specific goal-relevant regions. Such a structure is well suited to hyperbolic geometry. Motivated by this observation, we propose ProWorld, a progress-aware hyperbolic visual world model. ProWorld leverages goal-conditioned progress order to organize visual latent-space dynamics, maintains directional progress within trajectories via hyperbolic entailment learning, and mitigates progress ambiguity among locally similar future states via hyperbolic future discrimination. Furthermore, we design a progress-aware planning objective that scores candidate rollouts by jointly considering proximity to the goal and sustained progress across intermediate states. Experiments on four visual goal-reaching tasks demonstrate that ProWorld achieves an average absolute success-rate gain of 9.67 over LeWM. The code will be released after the paper is accepted.

Deep search agents answer difficult information-seeking questions by iteratively issuing search queries to gather supporting evidence, but it remains unclear whether and how greater search effort leads to better answers. We study these questions through a trajectory-level diagnosis of long-horizon search agents. Using human-annotated document-level relevance judgments, we evaluate the evidence retrieved at each search step and separate two stages of agent behavior: what evidence an agent retrieves and how effectively it uses that evidence. This distinction further allows us to decompose failures into retrieval gaps, where the necessary evidence is never found, and utilization gaps, where relevant evidence is retrieved but not used correctly. With the retrieval model and evaluation harness held fixed, we compare six agents on BrowseComp-Plus and further validate our findings on BrowseComp with an open-web search API. Across settings, we find that search effort and answer quality are only weakly aligned. Answer accuracy is better correlated with the quality of retrieved evidence, especially cumulative retrieval recall, than with the number of searches or the amount of context consumed. Useful evidence often appears early in the trajectory, yet agents tend to continue searching, producing a long tail of low-yield retrieval steps. At the query level, exploratory reformulations remain useful, but the best-performing agents issue far fewer redundant queries. Overall, by systematically characterizing the search behavior and failure modes of long-horizon search agents, this work points to practical directions for building better deep research systems, including stronger query formulation, more effective evidence selection and context management, and stopping criteria based on whether sufficient supporting evidence has been retrieved.

Multimodal large language models (MLLMs) have advanced visual understanding and reasoning, yet their static parametric knowledge limits their ability to address knowledge-intensive and dynamically evolving open-world problems. To move beyond this limitation, multimodal deep search has emerged as a key direction for open-world information access, evolving from single-turn factual retrieval toward long-horizon, multi-turn search guided by visual evidence. However, existing methods typically confine vision to the input or answer stage, overlooking its role in intermediate reasoning, and lack designs tailored to long-horizon interaction. Consequently, visual evidence rarely drives continued retrieval, constraining both interaction depth and reasoning span. To address these limitations, we propose DeepVoyager-VL, a long-horizon multimodal deep-search framework for vision-in-the-loop search. Specifically, we construct a multimodal event graph to drive data synthesis, yielding problems with intermediate visual dependencies and long reasoning chains. We then design an agent framework for active visual acquisition and on-demand image loading. Finally, we fine-tune models on the synthesized data without reinforcement learning. Extensive experiments across ten multimodal search benchmarks demonstrate the effectiveness of our method.

Long-horizon tasks require agents to maintain coherent state and goals across nested and branching work. We call this capability goal-directed execution (GDE): the repeated application of four behaviors, namely selecting goals, constructing task-relevant state, maintaining fidelity to higher-level objectives, and verifying completion against the environment. We hypothesize that long-horizon post-training strengthens these behaviors across domains. We test this by post-training Qwen3.5-122B-A10B on 363 Long-Horizon Multi-Tool Agent (LHMTA) tasks drawn from office workflows. The collection contained no software-engineering tasks, yet the model's pass@1 improved by 5.8 points on SWE-Bench Pro. Matched trajectory analysis shows gains in all four GDE behaviors in both office workflows and software repositories. Aggregate SWE-Bench Pro statistics showed related changes in information gathering, implementation, and verification. Together, the results support a behavioral interpretation in which long-horizon post-training changed how the model organized and applied knowledge across tasks, with effects extending beyond the training domain.

Machine-learning benchmarks often pair a label that aggregates a long temporal horizon with input observed through one or a few short windows. Their apparent performance ceiling may therefore be an acquisition-protocol ceiling rather than a model-capacity ceiling. We study labels of the form $Θ_{g,T}=T^{-1}\int_0^T g\{Z(t)\}\,\mathrm{d}t$ when the latent Gaussian process contains both a stable individual trait and a correlated within-individual state. An exact protocol-conditioned Bayes-risk identity provides a common tool. First, we decompose label variance into an $O(1)$ trait component and an $O(T^{-1})$ state component, explaining why a snapshot can retain cross-sectional predictability while poorly tracking within-person change. Second, we derive task-dependent effective temporal spans: mean labels depend on the ordinary correlation time, whereas occupation-time labels depend on an entire spectrum of higher-order correlation times. Third, state-driven occupation-label variance is maximal when the stable trait lies at the threshold; window efficiency decays much more slowly away from that boundary. Under an equal segment budget, exact risks and Monte Carlo experiments show that repeated segments at one time rapidly saturate, whereas temporally dispersed observations continue to increase state explainability. The trait ceiling uses quantities available from ordinary test-retest data; only the state ceiling requires short-lag temporal calibration. The results distinguish architectural limits from protocol limits and show that the label, rather than duration or segment count alone, defines the relevant timescale.

Agents are increasingly expected to act not only as task executors, but also as decision-makers on behalf of human users. This shift requires agents to accumulate evidence over long horizons, interpret implicit user preferences, and compare multiple candidates under partial observations. In this work, we propose DunphyBench, a new benchmark for evaluating agents on long-horizon human-centered embodied decision-making, where the agent must navigate through multiple embodied housing environments and make decisions that align with multi-dimensional human preferences. Unlike standard embodied reasoning tasks that often focus on procedural planning or immediate goal completion, our setting requires agents to integrate multimodal, multi-source input into coherent knowledge that supports complex reasoning across long horizon. The evaluation results reveal that there is a substantial gap between current agents and human performance. Furthermore, our diagnosis of state-of-the-art VLM-driven agents reveals that memory management is one of the bottlenecks, where raw multimodal history introduces noise that hinders decision quality. Motivated by this finding, we design MeMento, a preference-conditioned multimodal memory compressor that selectively compresses decision-relevant information from long-horizon history based on user preferences with a fixed set of memory tokens. Experiments show that MeMento helps VLM-driven agents improve accuracy by 7.18%, while reducing memory usage by 85.38% compared to the strongest baseline.

Multimodal Continual Instruction Tuning (MCIT) enables multimodal large language models to acquire new tasks sequentially while retaining previously learned capabilities. Many recent methods maintain task-specific LoRA experts and route each input to one or more experts at inference. Yet the task-identification problem underlying expert routing remains under-explored. We show that routing is nearly saturated on widely used MCIT benchmarks. Textual fingerprints that leak task identity and short 4--10-task sequences with few competing experts jointly obscure the long-horizon routing problem. To expose this challenge, we introduce FLEX (Fingerprint-reduced Long-horizon Expert eXamination), a 34-task long-horizon MCIT benchmark with weakened textual fingerprints. FLEX groups tasks with similar instruction and answer formats but diverse visual and knowledge domains, normalizes their outer templates, and evaluates routing over a substantially larger expert pool. Crucially, we formulate progressive-LoRA routing as soft task-as-class Multimodal Class-Incremental Learning (MCIL): each task defines an incremental routing class, whose complete score distribution supplies the LoRA mixture weights, with hard routing as a discrete special case. FLEX exposes this expanding task-identification challenge, while the MCIL formulation provides a principled interface for transferring CIL methods to expert routing. We instantiate PureLoRA as a controlled baseline and adapt four CIL methods to four MCIT frameworks without modifying their LoRA experts or generation pipelines. Our plug-in routers improve strict LoRA matching by up to 16.3 percentage points and overall MacroScore by up to 4.6 points. Code is available at: https://github.com/RINC-CL/FLEX

The continued development of LLMs toward persistent and adaptive intelligence increasingly requires long-term memory mechanisms that preserve and reuse information across interactions. Existing memory systems either compress and structure histories for efficient access or perform deep research over broader trajectories. The former lowers online cost but may omit temporal, causal, or cross-step dependencies, while the latter improves evidence coverage at substantial latency and inference cost. This raises a key question: can a memory system achieve strong answer quality while maintaining low online latency? We introduce Router-Mem, an evidence-conditioned progressive execution framework for long-horizon agent memory. Router-Mem first applies a shared low-cost retrieval prefix to obtain evidence. A lightweight sufficiency router then predicts whether the context supports early termination, which enable a single-token decision at inference time. It is trained with evidence-level supervision and rationale-conditioned representation distillation. When evidence is insufficient, Router-Mem reuses retrieval hits to expand memory blocks and perform deeper analysis and aggregation. Experiments on AMA-Bench and BEAM show that Router-Mem achieves 55.17% and 38.77% score while reducing average inference time by 27.3% and 25.5% compared with full memory execution.

Conversational investment advisors influence not only what users know, but also how they make subsequent decisions as market conditions evolve. Existing evaluations primarily assess response quality or observed outcomes, leaving the long-horizon pathway from advisor language to investor behavior difficult to audit. We introduce ShiJianBench, an offline framework for evaluating conversational investment advisors through matched investor trajectories under fixed historical market feedback. At its core is a multi-agent investor simulator with explicit evolving state variables, motive-driven deliberation, long-term memory, and dialogue-grounded updates. The simulator is calibrated against aggregate behavioral patterns from 7,199 real users, and advisor policies are evaluated using separate investor-side, service-side, and content-side metrics under a hard compliance gate. Experiments on Chinese fund-market traces from 2021 to 2026 identify a stable leading group of LLM advisors that combines substantially stronger personalized content with competitive investor-side trajectory outcomes. These results reveal a systematic distinction between producing a high-quality response and delivering an effective long-horizon intervention, motivating trajectory-aware evaluation of conversational advisors.

Lifelong LLM agents increasingly adapt through external learning states that store past interactions as retrievable memories or reusable skills, yet existing benchmarks rarely account for how the path of accumulated experience shapes what agents transfer and retain. In this work, we establish PATH-Bench, a benchmark for path-dependent evaluation of lifelong agents. PATH-Bench estimates directed task relationships via multi-model in-context learning, constructs probe-centered sequences with controlled helpful and interfering histories, and repeatedly evaluates probe tasks to measure average performance, forward transfer, backward transfer, and forgetting. We evaluate eight representative agents on single-turn code generation and multi-turn tool-use tasks under positive- and negative-dominant histories. Benchmark results show that experience utility depends jointly on how experience is represented and on the task's interaction structure, that strong transfer does not ensure retention, and that later experience can reshape gains acquired earlier in the learning path. Based on these findings, we propose Selective Experience Use (SEU), an agent harness that regulates how path-accumulated experience influences each new task, admitting helpful items while filtering out potential interference. SEU consistently reduces forgetting while improving forward transfer in the majority of settings. The PATH-Bench provides both a controlled evaluation framework and actionable guidance for designing more selective and robust lifelong agents.

For reinforcement learning (RL) in self-contained environments, a policy can get rewards by exploiting environment-specific regularities (tool schemas, grader parsing, task templates) rather than by acquiring transferable skill, and an in-distribution holdout shares those regularities. We argue that the discriminating question is behavioral, namely how a trained agent acts, and that cross-benchmark transfer is the right place to look for it. We post-train an open-weight mixture-of-experts model (Qwen3.5-122B-A10B) on 363 long-horizon Model Context Protocol (MCP) tasks across 27 categories, using a two-stage SFT-then-RL pipeline. Toolathlon performance informed the initial base-family and SFT-teacher choices, but no external-benchmark task or grader entered training and no external score informed the reward, training hyperparameters, trained-checkpoint selection, or stopping. At greedy pass@1, the trained model improves over the base on five reported external evaluations: Toolathlon (+9.6 pp), $τ^2$-Bench (+5.3 pp), BFCL-V4 (+3.5 pp), SWE-Bench Pro (+5.8 pp), and Terminal-Bench 2 (+2.8 pp). Both software-engineering benchmarks improve despite the training collection containing no software-engineering tasks. An exploratory paired-trajectory analysis identifies four recurring behavioral differences (more careful local-goal formation, building goal-relevant working state, keeping parent goals stable through local repairs, and verifying completion) that appear in analogous forms across office workflows and code. These results provide descriptive evidence that long-horizon multi-tool post-training can change ways of working that transfer beyond its training domain.

Symbolic Regression (SR) aims to discover analytical equations from observational data and plays a central role in scientific modeling. While recent Large Language Model (LLM) based approaches show promise, they face two limitations. First, they lack data analysis mechanisms for uncovering variable dependencies, which reduces the efficiency of equation discovery. Second, most methods rely on single-objective evaluation focused solely on fitting error. This neglect of structural complexity and generalization often causes models to converge prematurely to local optima, limiting their ability to explore the broader equation space. We propose Multi-Objective Tool-augmented Symbolic Regression (MOT-SR), a unified framework that integrates external analytical tools to extract structural priors and guide equation generation, while jointly optimizing for accuracy, complexity, and generalization via a multi-objective evaluation module that maintains a dynamic Pareto front. MOT-SR employs two collaborative LLM modules: a Meta Strategy Generator, which selects tools and synthesizes structural optimization strategies based on Pareto-optimal equations, and an Equation Generator, which produces new candidate equations accordingly. The system operates in a closed-loop manner, continuously refining both strategies and equation structures. Across 40 standard tasks, MOT-SR outperforms existing SR methods in accuracy, generalization, and efficiency. We further validate MOT-SR on extreme mass-ratio inspiral (EMRI) orbital modeling, an important problem in space-based gravitational-wave astronomy where small local errors can accumulate substantially over long-term evolution. The discovered interpretable correction achieves the lowest trajectory-level integration error on held-out configurations. These results demonstrate the potential of MOT-SR to enable reliable modeling of long-horizon scientific dynamics.

Neural operators provide fast surrogates for time-dependent partial differential equations (PDEs) by applying a learned evolution operator recursively to its own predictions, but this autoregressive rollout feeds every prediction error back as input, so local errors accumulate. Existing rollout-training strategies reduce the mismatch between training inputs and self-generated states, yet their supervision still measures only the absolute discrepancy from the ground-truth trajectory. Such supervision is therefore uninformative about whether the operator has overcome the long-horizon failure behaviors it exhibited earlier during optimization. We propose history-enriched rollout training (HERO), which augments conventional absolute trajectory supervision with relative supervision derived from the model's optimization history. HERO ranks detached candidate rollouts from a periodically refreshed lagged operator, the current model, and a perturbed input by rollout error, spectral discrepancy, energy drift, and error growth, and selects the strongest failure trajectory as reference. This reference enters a margin-based objective as a fixed comparison baseline, inducing a bounded, sample-dependent reweighting of the ground-truth rollout gradient rather than an independent gradient direction, which we further analyze theoretically. Experiments on nine PDE benchmarks with spectral and attention-based backbones show that HERO consistently improves long-horizon accuracy, stable rollout length, and out-of-distribution robustness at no inference-time cost. These results indicate that history-enriched relative supervision is effective for stabilizing long-horizon autoregressive prediction.

Neural operators have shown strong potential for learning solution operators of partial differential equations (PDEs). However, long-horizon autoregressive prediction remains challenging: local errors accumulate as spectral inconsistency, phase misalignment, or mean drift. Existing methods mainly improve state representations and operator backbones, while leaving the repeatedly applied latent transition increment weakly structured, allowing spectral errors and unstable channel couplings to accumulate during rollout. To address these issues, we propose a geometry-aware incremental neural operator (GeoIncNO) for stable long-horizon PDE prediction. GeoIncNO predicts latent increments for residual advancement and uses lightweight low-rank projectors to regulate channel coupling within active frequency bands derived from the increment spectral energy distribution. To reduce physical-space reconstruction errors, GeoIncNO further introduces a mean--fluctuation decoupled reconstruction mechanism, where stable mean structures and dynamic fluctuations are fused separately, and phase correction is applied only to the zero-mean fluctuation component. Extensive experiments on six PDE benchmarks, covering 1D, 2D, and 3D dynamical systems, show that GeoIncNO achieves consistently strong prediction accuracy, improved rollout stability, and better spectral fidelity compared with competitive neural-operator baselines.

As AI companions increasingly mediate repeated social interaction, users may rely on a stable role and shared history, yet locally acceptable replies do not ensure that either persists. We study two observable long-horizon failures: 'persona collapse', the loss of a deployed role, boundaries, values, or style, and 'behavioral drift', the gradual or recurrent erosion of those properties. We introduce ANCHOR, a controlled synthetic audit that separately measures persona enactment and trajectory recall. The study contains 2,008 conversations spanning 27 personas, nine interaction schedules, three generated memory settings, and four evaluated models. The Identity Probe combines a sealed 102-item questionnaire with turn-level judgments, while the Trajectory Probe scores 110 calibrated counterfactual questions from 35 conversation banks. Our results show that no evaluated model and configuration reliably preserves either dimensions: trajectory accuracy averages only 44.4%, user-state recall remains near four-option chance, and no tested context condition or memory consistently resolves these failures. Questionnaire retention also varies by model and persona facet, disagrees with turn-level behavior, and is sensitive to evaluator choice. These results indicate that current systems do not yet reliably support long-horizon companion continuity and that audits must distinguish persona enactment, trajectory recall, evaluator provenance, and deployment context rather than collapse them into a single trust or stability score.

Understanding large codebases is a long-horizon task for Large Language Model (LLM) agents: answering a single question can require building and running the software, tracing execution across files, and synthesizing evidence over tens of minutes. On SWE-Atlas QnA, a benchmark of long-horizon questions over production repositories, a single Claude Code agent (Opus 4.6) resolves only 32.3% of tasks. Dividing the work among agents with clean contexts mitigates this limitation. However, the subtasks of code comprehension are interdependent. One agent's findings can rewrite another's task, so agents must coordinate during execution, not only at phase boundaries. Existing multi-agent systems support such exchange only between phases, through staged handoffs or synchronized rounds. Communication and work remain mutually exclusive. A discovery made mid-execution cannot be shared until the next boundary. We present AgentRadio, an asynchronous message-passing layer that equips coding-agent harnesses with three primitives: threads, messages, and waiting for mentions. The last runs as a background task, surfacing teammates' messages without interrupting foreground work, so each agent remains passively aware of its peers and folds new findings into its ongoing task. Under a five-phase protocol of division of labor and negotiation, four agents organized by AgentRadio resolve 62.1% of tasks, 29.8 points above a single agent and above Claude Code with the newer Opus 4.8 (57.2%). Rubric-level analysis shows the gain growing with task difficulty, consistent with mid-course correction as the underlying mechanism. Our code is available at https://github.com/Coral-Protocol/AgentRadio.

Many dexterous manipulation tasks require the object to remain securely held throughout the interaction. From the perspective of hand-object relational motion, such manipulation comprises four canonical skills: grasping, relocation, in-hand rotation, and in-hand translation. Human hands flexibly compose these skills to accomplish complex tasks. Existing approaches, however, model these skills separately with skill-specific action constraints, objectives, or even dedicated hand morphologies, which breaks the compatibility and continuity required for long-horizon composition. In this work, we present a unified framework that models all four skills in a single formulation that shares the same state and action spaces and a common objective structure. This formulation enables straightforward distillation of a single cross-skill policy that performs strongly on every skill, generalizes to unseen objects, stays robust to disturbances, and chains skills seamlessly into long-horizon manipulation. The framework also transfers effectively across different hand morphologies. Overall, our results suggest that different dexterous manipulation skills can be viewed as instantiations of a shared task formulation, revealing the intrinsic consistency across different behaviors.

Long-horizon steel-equipment inspection requires reasoning over heterogeneous records accumulated across repeated inspection cycles. Existing retrieval-augmented generation systems treat historical logs as a static corpus and retain records without estimating their diagnostic value, failing to report early risk. To this end, we propose ConMem, a contribution-aware memory framework for LLM-assisted equipment inspection, supporting a human-in-the-loop early-risk screening system. Specifically, our ConMem first segments inspection logs into functional evidence units, then estimates each memory unit's contribution to downstream diagnosis through a Shapley-style estimation, and finally retains high-value evidence under a constrained memory budget. In experiments, we evaluate ConMem on real-world dataset and ConMem achieves 76.0% QA accuracy, exceeding the strongest directly comparable baseline. Relative to the naive 8K-context LLM baselines, it reduces the average number of input tokens by 88.2% and response time by 86.6%. Ablation studies also show that the functional-role-aware segmentation and contribution-based valuation are helping prioritize weak degradation signals for targeted field inspection. Practical deployments further confirm that ConMem retains the weak early signal across three inspection cycles, providing an early-stage seal-wear alert targeted for on-site inspectors.

Large language models can contribute useful ideas to mathematical research, yet long-horizon proof attempts remain difficult to coordinate, evaluate, and reproduce. We present Albilich, an open-source agentic harness for autoresearch in mathematics that combines long-horizon reasoning, computer algebra systems (CAS), literature retrieval, and persistent SQLite-based context management. We evaluate Albilich on the RealMath benchmark (Zhang et al. 2025) and on open problems in group theory from the Kourovka Notebook (Khukhro and Mazurov 2026). It solved 10/10 problems on RealMath with CAS and 9/10 with no CAS. On the Kourovka problems, Albilich produced a counterexample to Problem 21.142 and a proof of a strengthening of Problem20.2. Anablation on Problem 17.91 demonstrates 32.0% token reduction when CAS is enabled. An ablation on Problem 21.142 demonstrates higher verifier-rejection rate and failure to synthesize proof routes in the absence of the advisor agent. These results support Albilich as a human-steerable, CAS-boosted environment for scalable AI-assisted mathematical research.

Long-horizon benchmarks often show that agents fail more as tasks become longer. This observation is useful for deployment, but it does not by itself explain why failure occurs. More stages create more opportunities for ordinary errors to compound; longer tasks may also contain harder individual decisions or become harder as conversation history, tool outputs, and environment changes accumulate. We use trajectory-induced degradation to mean this last possibility: earlier execution makes later work harder. When the harmful accumulation is specifically the text visible to the model, it is often called context rot. In this position paper, we argue that to claim a "long-horizon failure", benchmarks must compare actual full-task success against a baseline prediction built from short, individual stages. We call the log-ratio between this prediction and actual success the horizon residual. The comparison must use the same agent configuration and specify in advance how stages, checkpoints, information, and budgets will be chosen. The residual shows that the full rollout differs from the chosen baseline; targeted experiments are still needed to explain why.

Vision-language-action (VLA) policies commonly execute long-horizon mobile manipulation through open-loop action chunks, issuing multiple actions without receiving new high-level visual input. A committed chunk therefore implies how observations should evolve, but accidental deviations can violate this expectation while the remaining actions continue to propagate the error: commit-time policy confidence cannot react to a deviation that occurs after dispatch, and observation-only anomaly scores lack an action-conditioned reference for separating expected effects from unexplained changes. We propose CheckVLA, which verifies execution with a separately trained, frozen action-conditioned world model. A conformally calibrated risk threshold bounds the episode-level probability of an unnecessary first intervention and determines when to intervene, its exceedance controls how strongly the rewritten suffix retains the superseded chunk, latency-aware hard prefixing restricts replacement to actions that remain deployable, and an event-driven keyframe bank preserves evidence of prior progress across repairs. On RoboCasa365, under a common training recipe and a matched invocation budget, CheckVLA attains a 36.1% average success rate against 27.6% for periodic replanning (+8.5 points). At a matched 5% episode-level false-alarm target, action conditioning raises timely recall to 77.9%, against 48.6% for an observation-only control and 37.9% for an action-shuffled control. These simulation results support action-conditioned verification as a way to restore feedback during chunked execution while keeping the repair consistent with inference latency.

Latent world models support efficient model-predictive control by optimizing future control sequences in latent space and replanning in a receding-horizon manner. However, existing latent predictors often lack stable long-horizon rollout ability, and prediction accuracy alone does not ensure that rollouts remain responsive to the actions being planned. We identify Context Collapse, a failure mode in which autoregressive latent predictors maintain high similarity to future states while producing nearly indistinguishable futures under different action sequences. To address this issue, we propose ActSWM, an action-sensitive latent world model grounded in a transition-separation principle: a planning-useful latent dynamics model should keep alternative-action futures distinguishable and make the action associated with each local transition recoverable. Under this principle, action sensitivity is enforced as a constraint on latent rollouts rather than treated only as an auxiliary prediction target, encouraging predicted futures to preserve action-dependent differences over long horizons. Across step-drift analysis, closed-loop Minecraft planning, and cross-game local action recovery, ActSWM preserves larger action-dependent rollout gaps than existing baselines, improves task success in long-horizon interactive settings, and enables world-model-based action recovery from offline gameplay videos.

Clinical data-science agents must transform heterogeneous longitudinal records into auditable analyses, yet existing benchmarks largely isolate medical question answering, structured-table reasoning, or generic scientific repositories. We introduce CLINLENS, a benchmark of 200 executable tasks over five linked MIMIC resources spanning structured electronic health records, notes, electrocardiograms, chest radiographs, and echocardiograms. A 4 x 5 taxonomy crosses four patient-time scopes with five analysis capabilities. Program-first reverse synthesis pairs each bounded semi-raw package with an evaluator-private reference workflow and checks required artifacts, cohort and temporal semantics, and the final answer. On a fixed 126-task suite, the strongest of 24 standardized model-scaffold configurations achieves 56.3% scope-macro STRICTPASS despite 100% EXECSUCCESS. For reference, a separately configured coding agent solves 83 of 126 tasks, while five biomedical systems adapted to GPT-4o-mini reach at most 2.9% scope-macro STRICTPASS. These results expose a substantial gap between runnable submissions and correct clinical analyses.

Memory is essential for LLM agents to accumulate task experience and reuse task-specific execution strategies. However, real-world deployment over boundary-agnostic and evolving task streams exposes a fundamental stability-plasticity dilemma. External retrieval-based memory can rapidly absorb new evidence, but it often fails to internalize recurring execution patterns and incurs inference-time retrieval overhead. Parametric memory enables stable and efficient execution once learned, but typically relies on explicit task boundaries and fixed parameter budgets. Inspired by the human brain, which balances plasticity and stability through complementary episodic storage and gradual consolidation, we propose UniMem, a self-routing framework for autonomous memory management. UniMem uses learnable routing tokens as memory controllers, enabling adaptive coordination between complementary memory pathways: novel or sparse tasks are retained in an episodic buffer for retrieval-augmented execution, while recurring and reliable patterns are consolidated into expandable parametric memory. By decoupling task identification from task execution with routing tokens and parametric memory blocks, UniMem expands memory on demand without task labels during deployment or uncontrolled parameter growth. Experiments on long-horizon streaming task sequences show that UniMem consistently outperforms baselines while maintaining execution fidelity, achieving an average gain of 4.0 EM points across three backbone models.

Skills have become an important abstraction for enabling large language model (LLM) agents to reuse past experience in long-horizon interactive tasks. However, existing trajectory-to-skill methods often produce flat collections of high-level textual skills that are stored and retrieved independently, leaving skill relations underutilized and maintaining a gap between high-level skills and executable actions. In this paper, we propose HiSkill, a hierarchical skill graph framework that organizes interaction trajectories into a directed graph with skill nodes, AtomicOp nodes, and typed edges. Specifically, the graph connects reusable high-level skills with executable action templates, while also capturing decomposition, temporal transition, compatibility, support, and recovery relations among them. At inference time, HiSkill retrieves a compact task-relevant subgraph and performs subgraph-guided task execution, where a symbolic task state, an active skill, and the retrieved subgraph guide the LLM agent to switch skills, select AtomicOps, and ground executable actions iteratively. Experiments on three interactive environments show that HiSkill outperforms state-of-the-art baselines while reducing inference token consumption, demonstrating the effectiveness of bridging high-level skills and executable action grounding through a hierarchical skill graph. Our data and code is available at https://github.com/BUPT-GAMMA/HiSkill.

Agent harnesses have become the operational infrastructure of modern large language model agents, coordinating context, tools, verification, and execution control to translate latent model capability into reliable long-horizon behavior. However, reliable long-horizon behavior requires harness control to adapt to task demands, execution environments, and evolving execution states, whereas current harnesses predominantly rely on hand-crafted or globally fixed policies; this mismatch manifests as unnecessary computational overhead and, in adverse cases, reduced task success. To address this limitation, we formulate the task of enabling adaptive orchestration in harness systems as a causal learning problem and propose Counterfactual Harness Intervention Learning for Long-Horizon Agents (CHILL-Harness). CHILL-Harness intervenes at the orchestration layer to enable advantage-guided workflow adaptation, thereby improving reasoning and execution efficiency while preserving task performance. Specifically, we develop causal intervention effect learning as the effect-estimation component of CHILL-Harness to estimate intervention-relative workflow advantage from confidence-weighted execution evidence and identify advantageous workflow adaptations. We further introduce advantage-realizing causal orchestration as its realization component to adaptively allocate counterfactual reasoning and realize only workflow adjustments supported by sufficient expected advantage. Finally, we incorporate a success-preserving objective and advantage-margin authorization constraints into CHILL-Harness to promote reliable adaptation. Extensive experiments on heterogeneous long-horizon tasks spanning information seeking, software engineering, and terminal interaction show that CHILL-Harness consistently preserves or improves task success while substantially reducing token consumption and execution time.

Machine learning engineering (MLE) tasks require long-horizon decision making over iterative solution debugging and refinement, under expensive and feedback-driven environment interactions. Developing and training a monolithic agent for such tasks is fundamentally challenging, as it must simultaneously manage extremely long and noisy contexts, explore vast solution spaces, and remain effective under limited model capacity and computational budgets. To address these challenges, we propose Matryoshka Agent, a unified hierarchical agent framework for complex long-horizon tasks. Matryoshka Agent decomposes agentic problem solving into a coordinated hierarchy of decision making and execution: a high-level Orchestrator maintains compact, long-horizon exploration states and issues strategic instructions, while lower-level Sub-Agents execute concrete solution attempts through direct environment interaction, mediated by standardized Tool interface. This design decouples strategic exploration from costly execution, substantially reducing the burden of long-context reasoning and enabling efficient iterative refinement. We further develop an efficient training paradigm for Matryoshka Agent. Experimental results on a broad range of MLE tasks with diverse model types and scales demonstrate that Matryoshka Agent is an effective and scalable paradigm for long-horizon MLE tasks and complex agentic problem solving. Notably, Matryoshka Agent enables Qwen3-4B-Instruct to reach Orchestrator performance comparable to o4-mini. Applying Matryoshka Agent to Qwen3-30B-Coder results in at most 36.7% relative performance gain.

Long-horizon LLM agents accumulate reasoning traces, actions, and tool observations that can eventually exceed a model's fixed context window. Existing compaction methods address this limitation by discarding, summarizing, or retrieving earlier information, but they may remove task-critical details or fail to recover them reliably. We propose ARC (Addressable Recall Compaction), a context-management framework that separates archival storage from active-context presentation. ARC stores tool observations in an append-only, ID-addressable log and replaces older observations with compact citations when compaction is required. The agent can subsequently use these identifiers to request stored content without re-executing the corresponding tools or depending solely on similarity-based retrieval. We evaluate ARC using Qwen3-8B with a 16k context window and Qwen3-32B with a 32k context window. On the Needle-in-a-Haystack evaluation, ARC achieves an average exact-answer accuracy of 99.40%, compared with 88.12% for the best-performing baseline in our evaluation. ARC also reduces estimated serving time and HBM traffic under our hardware-cost model. On the LongBench-v2 Hard subset, ARC obtains an average accuracy of 29.97%, compared with 28.25% for the best-performing baseline. These results indicate that explicit, address-based recall can improve information retention and serving efficiency relative to the evaluated context-management baselines under the tested settings.

AI "personality clones" force a re-examination of personal identity in operational terms. Setting aside the hard problem of consciousness, we approach identity through the indiscernibility of manifestations, as assessed by an observer over a duration. We distinguish three criteria that "identity" conflates: fidelity to a target person, generic human-likeness, and individuality. We propose a six-term factorization of observed identity (substrate, dispositions, memory, update dynamics, context, exogenous contingencies), with a state-space formulation. Indiscernibility is defined as one minus a judge's distinguishing advantage, and the factorization's coefficients become local sensitivities estimable by randomized ablation. The central claim is a conditional conjecture: given hypotheses about the agent's information on its own persistence and about consequences bearing on its own stakes, versionability tends to degrade long-horizon indiscernibility. An analogy with lambda-calculus, linear typing, and bisimulation clarifies what linearity does and does not establish. Between product-clone and individual we identify a third object, the delegate: a task-limited, bounded-lifespan partial clone ending in a bandwidth-limited testament. We map the empirical literature onto the three criteria, propose an experimental program, and argue that the correct long-horizon criterion is not trajectory fidelity but climate fidelity: matching the conditional distribution of a person's possible responses. The best clone is the one that diverges from the original as the original would have diverged from itself.

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. To address this challenge, we introduce a unified and controlled multi-turn environment that enables precise control. It allows systematically study long-horizon planning across three stages. (1) Planning ability acquisition during pre-training. We study data format, distribution, and quality. Explicit world model construction through CoT state transition modeling yields stronger long-horizon generalization. Atomic skills alone are insufficient for compositional generalization, whereas a litte long-horizon data works. Moreover, suboptimal trajectories severely impair performance because errors amplify over long horizons. (2) Planning ability shaping via GRPO and OPD post-training. Through mutual information, we distinguish general planning patterns from task-specific planning knowledge. For planning patterns, we identify three application regions of post-training: unnecessary, effective, and unsupported. OPD has a broader effective region than GRPO under low-quality and long-horizon settings, as it provides more consistent update directions. For planning knowledge, distilling unseen procedures from a teacher with different knowledge may impair student's prior world modeling without fully establishing new knowledge. (3) Planning ability integration through MOPD post-training. We show that multi-teacher on-policy distillation (MOPD) integrates capabilities by converging to shared planning-pattern across environments. Compatible patterns enable cross-environment generalization, partially shared patterns support continual learning, while completely conflicting patterns cause severe interference.

Long-horizon web agents often go off track before final failure: a trajectory can remain locally plausible even after the current state, reused skill, or plan assumption no longer supports the user instruction. Existing agents can plan, reflect, or reuse experience, but their plans rarely specify the evidence under which an active step should still be trusted. We propose FCPAgent, a falsifiable commitment planning framework for robust long-horizon web agents. FCPAgent represents each plan step as a Falsifiable Commitment Unit (FCU): a subgoal grounded in a reusable skill, together with confirming evidence, falsifying evidence, and a confidence score. Execution is organized as a plan-test-repair loop. The hybrid commitment testing module checks candidate actions before they modify the browser and checks observations after execution; for efficiency, it combines lightweight evidence matching with LLM-based diagnostic verification. When evidence falsifies a commitment, scope-aware repair localizes the contradiction to the execution, skill, or planning level and revises the smallest adequate part. On WebArena, FCPAgent achieves a 13.8% relative improvement in average success over the strongest baseline, with especially large gains on long-horizon tasks.

Agentic tasks are inherently long-horizon and multi-turn, constantly accumulating context through interactions with the environment. Existing context compression methods inevitably incur information loss and are triggered by rigid heuristic rules, leaving them misaligned with the agent's evolving reasoning focus. We propose Agentic Context Management (ACM), a framework that equips agents with purpose-built context editing tools for lossless context management. Inspired by the interaction between short-term and long-term human memory, the agent autonomously decides when to compress its context, offloads discarded content to an external memory system, and queries it on demand for later retrieval. Building on this framework, we further develop a post-training pipeline that constructs high-quality demonstrations of context management and improves model performance on both agentic search and coding tasks. Further analysis reveals that effective context management reduces peak token pressure, enables extended explorations, and yields more consistent solutions across independent trials. Code, data, and model checkpoints are available at https://github.com/lixiaochuan2020/agentic-context-management.

Exploration in sparse-reward long-horizon tasks poses significant challenges for reinforcement learning. To address these challenges, we propose a two-level Hierarchical Reinforcement Learning (HRL) framework. The first level handles high-level strategic planning, while the low-level uses the continuous-control Soft Actor-Critic (SAC) algorithm, and they utilize entropy-regularized policy optimization. The proposed framework was trained and evaluated using the Search-and-Rescue-2 (SAR-2) dataset. HRL-SAC effectively addresses sparse-reward long-horizon search problems characterized by delayed rewards and continuous control, and its outperforming the flat SAC baseline reinforcement learning in terms of success rates, coverage efficiency, and convergence. These findings indicate that hierarchical entropy-regularized policies are a promising solution to tackle long-horizon sparse-reward reinforcement learning tasks.

Large language models (LLMs) can summarize heterogeneous user evidence in natural language, but current LLM recommenders often collapse enduring preferences, transient intent, and exposure-induced behavior into one profile. This makes recommendation vulnerable to feedback loops: repeated exposure is mistaken for preference, immediate clicks dominate delayed satisfaction, and fluent explanations need not reflect the ranking decision. We propose our method, a model-agnostic framework for long-horizon recommendation. Our method uses a frozen multimodal language model to convert item content and feedback into evidence-grounded semantic atoms, then maintains separate short-term, long-term, and exposure memories. Propensity-weighted updates reduce policy-induced exposure bias, while a conservative offline critic reranks candidates for delayed satisfaction under a behavior-support constraint. Explanations use only influential evidence atoms and are checked by counterfactual deletion. We provide an identification result and evaluate the framework in e-commerce-like, news-like, and short-video-like environments. Across ten seeds, our method improves discounted long-term value over the strongest alternative by 6.1%, 7.6%, and 6.7%, respectively. Twenty-seed paired ablations show significant value drops after removing propensity correction (0.739 +/- 0.191) or conservative support regularization (0.523 +/- 0.234). A frozen instruction language model also more than doubles semantic-atom NDCG over TF-IDF on a held-out paraphrase benchmark.

Long-horizon manipulation tasks pose significant challenges for reinforcement learning due to sparse reward signals and long horizons. Automatic curriculum learning (ACL) has been proposed to tackle these challenges by progressively training agents on a sequence of tasks, from easier to more difficult. However, the success of ACL depends heavily on task-dependent specifications-such as well-defined task parameter spaces and difficulty measures-which are often manually crafted and difficult to generalize across diverse tasks. Recent advances in large language models (LLMs) offer a promising alternative by enabling the decomposition of complex tasks into meaningful subtasks using the LLMs' web-scale common-sense knowledge. This decomposition can provide a natural curriculum structure for efficient learning of long-horizon tasks. However, existing LLM-based methods typically rely on hand-designed dense reward functions to learn each subtask, which can introduce bias and still requires significant human supervision. In this work, we propose LLM-enhanced automatic curriculum learning (LEACL), a framework that integrates LLMs and ACL to address these limitations. Specifically, LLMs are used to both decompose tasks into subtasks and to generate task-dependent specifications for each subtask. These specifications are then used by ACL algorithms to guide learning using only sparse reward signals, eliminating the need for dense reward design. We evaluate LEACL on five long-horizon manipulation tasks from the LIBERO benchmark. LEACL achieves better asymptotic performance in terms of the success rates compared to human-designed dense rewards.

Language agents can now interact fluently with users in software, but robots still struggle to bring comparable interaction to physical tasks. Current robot-control paradigms, including vision-language-action policies and world-model-based planners, are mainly optimized for instruction execution, leaving users with little visibility into why an action is chosen and few mechanisms to redirect, correct, or teach the robot through interaction. To solve this problem, we present the World-Cognition Model (WCM), a human-centered embodied agent built on the SLAK architecture (Sensing, Logic, Action, and Knowledge) and an asynchronous runtime. SLAK separates perception, reasoning, control, and memory, while the runtime allows reasoning, dialogue, and execution to proceed concurrently. WCM further introduces a human-in-the-loop teaching mode that enables users to interactively teach the robot difficult or long-horizon tasks. Teaching episodes and autonomous task rollouts are refined into chain-of-thought supervision to continually improve the model. WCM achieves a 73.8% average success rate across nine real-world human-robot interaction tasks, including tasks held out from CoT fine-tuning and a long-horizon task learned through teaching.

Vision-and-Language Navigation (VLN) enables embodied agents to follow natural-language instructions. However, route-level instructions commonly encode spatial priors, such as orientation, distance, and layout, that are not explicitly available from onboard sensing at deployment in open, GPS-denied environments. Benchmark performance under such interfaces therefore jointly reflects visual navigation ability and the use of route structure explicitly supplied by the task description. As a complementary formulation, we propose Vision-Only Long-Horizon Navigation (VoLN), which shifts route-relevant information from externally supplied instructions and global guidance to locally observable in-scene cues. In VoLN, goal views specify the destination, while route-relevant information is available only through locally observable in-scene cues that the agent must detect, interpret, and select online. We instantiate VoLN for aerial navigation through VoLN-UAV, a 7,210-episode benchmark that combines long-horizon goal-directed flight, continuous 3D motion, large viewpoint changes, and context-dependent beacon selection. We further provide VoLN-MLLM as an initial reference baseline. It aligns self-supervised visual features with a structured semantic space and predicts short-horizon waypoint segments from observation history, goal views, retrieved visual--semantic tokens, and proprioception. On the five-environment Test-Unseen split, it obtains success rates of 7.4%, 4.5%, and 1.8% on Easy, Normal, and Hard episodes, respectively. These results provide an initial evaluation of VoLN and reveal substantial remaining challenges in long-horizon evidence integration, cross-view goal matching, and closed-loop stability. Project page: https://admire-ljb.github.io/VoLN-UAV/

Offline goal-conditioned reinforcement learning (RL) holds the promise of learning general-purpose policies from static datasets. However, scaling these methods to long-horizon tasks remains a challenge due to the curse of horizon, where value estimation errors can compound through long chains of bootstrapped Bellman backups. Existing hierarchical approaches mitigate this by decomposing tasks into subgoals, yet they often rely on low-level controllers that suffer from myopic execution and biased value estimates. In this work, we propose Hierarchical Implicit Q-Chunking (HiQC), an offline goal-conditioned RL algorithm that combines high-level latent planning with low-level action chunking. By conditioning the low-level critic on temporally extended action sequences, HiQC enables unbiased k-step value backups, compressing the horizon at both the planning and execution levels. We theoretically demonstrate that this dual decomposition results in a tighter bound on value error under a bounded per-backup error model compared to standard hierarchy or flat chunking alone. Empirically, HiQC achieves the highest aggregate performance among the compared methods on the OGBench suite, with its largest gains on long-horizon navigation tasks such as humanoid-giant.

Long-horizon tasks require sustained perception, reasoning, and exploration, and are a persistent challenge for large language model (LLM) agents. This gap is reflected in their limited performance on continual learning benchmarks such as ARC-AGI-3, especially when models are evaluated out of the box. Various agent harnesses have been proposed to close this gap, and each commits to a strategy for handling long sequences of observations, i.e., what information to save from the environment and how to load it into model context, a choice we argue is particularly consequential. Existing methods for context management face a significant tradeoff, as preserving more information makes retrieving relevant details less tractable. We propose PRO-LONG, a minimal context management framework built around programmatic memory for LLM agents in long-horizon, exploratory settings. PRO-LONG addresses the tradeoff by keeping a complete, structured interaction log and capitalizing on recent progress in coding agents to search this history efficiently. On the full ARC-AGI-3 public game set, PRO-LONG improves over a base coding agent by an average of 18.0 percentage points across frontier models, and matches or exceeds state-of-the-art specialized harnesses (up to 76.1% pass@1) while using 4.2-5.8x fewer tokens. With Fable 5, PRO-LONG achieves 97.4% best@2 at a total cost of $1,750. Relevant code and logs are available at https://github.com/alexisfox7/PRO-LONG.

We present Solar Open 2, a 250B-A15B Mixture-of-Experts language model built for long-horizon agentic tasks, scaled up from Solar Open 1 (Solar Open 100B). To hold entire agent trajectories in a single context, Solar Open 2 reaches a 1M-token window through a hybrid attention stack that interleaves one softmax layer among every three linear-attention layers, using no positional encoding and a gated delta rule extended to negative eigenvalues. To train at this scale under a fixed compute budget, we make training efficient in two ways: a stronger starting point, and higher-value data. For the starting point, we initialize Solar Open 2 from Solar Open 1, transferring the 5.69B-parameter shared skeleton that survives the architectural change and learning everything else through full pre-training. For the data, we curate for value per token: quality- and rarity-aware data curation and mixture-ratio optimization refine a 20T pool into a 10T mixture that, at equal token budget, outperforms the Solar Open 1 recipe. To build its agent skills, we train twelve domain specialists across purpose-built scenarios, then consolidate them into a single model by Multi-teacher On-Policy Distillation (MOPD). Against comparably sized open-weight models on English benchmarks, Solar Open 2 leads on MMLU-Pro, LiveCodeBench, and the APEX-Agents agentic suite, and stays competitive with the strongest (DeepSeek-V4-Flash and MiMo-V2.5) elsewhere. On Korean benchmarks, Solar Open 2 records the highest average of any model compared, including fast-tier closed APIs, and on Ko-GDPval, an in-house Korean officework-agent benchmark, it is competitive with DeepSeek-V4-Pro (1.6T) at less than a sixth of its size.

We study horizon-free regret minimization for finite-horizon time-homogeneous tabular Markov decision processes with $S$ states, $A$ actions, horizon $H$, and per-trajectory total reward bounded by $1$. We propose a new algorithm and prove a regret upper bound $$\tilde O(\sqrt{SAK}+S^8A^3)$$ with failure probability $δ$, where $K$ is the number of episodes and $\tilde O(\cdot)$ hides $\mathsf{poly}\log(S,A,K,1/δ)$. Thus, the regret is $H$-free and asymptotically optimal, matching the contextual-bandit lower bound $Ω(\sqrt{SAK})$ up to logarithmic factors. This completely removes the $\log H$ dependence from the previous $\tilde O(\sqrt{SAK\log H}+S^2A\log H)$ guarantee of Zhang et al. (2021), and drastically improves the prior best horizon-free regret $\tilde O(\sqrt{S^9A^3K})$ of Zhang et al. (2022) asymptotically. The main technical difficulty is that the optimal value functions $\{V_h^*\}_{h=1}^H$ are time-inhomogeneous even though the transition kernel is time-homogeneous. A direct union bound over all value functions typically incurs an additional $\min\{\log H,S\}$ factor. We avoid this factor by (i) exploiting the monotonicity of $V_h^*$ in $h$ and (ii) non-trivially projecting the value functions onto an $S$-dimensional grid. Our analysis relies on three additional ingredients. First, we introduce a horizon-truncation argument that enables reward-based exploration and removes the cost of a separate reward-free exploration phase. Second, we design a cutting bonus that preserves both optimism and the monotonicity needed for planning. Third, we prove a new bound on total deviation for time-homogeneous MDPs, which controls the clipped variance terms in the cutting bonus with adjustable polynomial dependence on $S$ and without any dependence on $H$. Together, these tools yield an asymptotically optimal horizon-free regret guarantee.

Do Large Language Models (LLMs) possess genuine structural reasoning, or merely rely on surface-level pattern matching? The financial domain, demanding numerical precision and multi-step logic over long contexts, is an ideal testbed. Existing benchmarks fail to capture real-world industrial complexity, predominantly relying on multiple-choice questions or single-hop QA over cropped tables while ignoring intricate cross-statement dynamics and temporal de-cumulation. To bridge this gap, we introduce FinIndices, a large-scale benchmark evaluating data-processing fidelity over uncropped financial statements (up to 32K tokens). Utilizing an automated synthesis pipeline with adversarial traps, FinIndices encompasses Single-Index computation and Table-Index tabulation to test complex domain, temporal, and caliber reasoning. Our evaluation reveals two severe LLM vulnerabilities. First, a "Knowledge Bottleneck": despite memorizing formulas during pre-training, models demonstrate fragile pattern matching. Removing explicit formula hints causes performance to collapse (e.g., Gemini-3.1-Pro drops from 70.70% to 38.22% on table tasks), exposing fatal flaws in temporal de-cumulation and stock-flow caliber mismatch. Second, a "Structural Bottleneck": the intense cognitive load of generating multi-metric, multi-period tables actively drains reasoning capacity. Under structural pressure, LLMs that flawlessly execute isolated derivations regress to shallow heuristics, such as fetching incorrect adjacent columns or substituting deep accounting adjustments with lazy literal arithmetic. Finally, Supervised Fine-Tuning (SFT) yields substantial zero-hint gains (+8.54% Single, +3.82% Table), validating that structured logic can be partially restored via data-centric alignment.

Group-based policy optimization has been increasingly used to train large language model (LLM) agents from sparse outcome rewards by comparing trajectories or steps within a group. However, on difficult long-horizon tasks, this comparison can suffer from a sampling imbalance: repeated or low-effect actions dominate the high-probability region of the policy while useful state-changing actions remain under-sampled. This imbalance produces many all-failed rollout groups, where outcome rewards provide no direction for correcting the policy. Together, these effects can form a self-reinforcing credit trap: failure-dominated sampling yields no outcome-based correction, allowing repeated low-effect actions to persist. To break this loop, we propose Progress-conditioned Group Policy Optimization (ProGPO), which uses first-visit observation coverage only when all samples in a group receive zero outcome reward. Specifically, within such groups, ProGPO assigns higher relative advantages to trajectories or steps that visit more new states since reaching new observations is a prerequisite for task success. Experiments on two challenging agentic benchmarks, ALFWorld and WebShop with Qwen2.5-1.5/7B-Instruct, show that ProGPO consistently improves over group-based baselines, with particularly large gains on hard tasks.

Long-horizon robotic tasks require diverse capabilities that no single policy can reliably provide. Heterogeneous policies offer complementary strengths, but orchestrating them requires reasoning over uncertain capability boundaries and cross-policy distribution mismatch, which are largely overlooked by existing planning methods built on homogeneous, predefined skills with fixed applicability. We propose RoboHarness, a unified framework that encapsulates independently developed robot control systems as reusable agentic skills. Although instantiated in this work with VLAs, RL policies, and task-and-motion planning (TAMP) systems, RoboHarness is designed as a general framework compatible with a broader range of robot policies, such as navigation policies, model predictive controllers, and world-action models. RoboHarness uses multi-modal execution memory and online evidence to characterize policy capability boundaries for capability-aware decomposition and routing. To stabilize policy handoffs, its Memory Bridge retrieves execution trajectories associated with the next policy, estimates its in-distribution state region, and guides the robot toward that region without joint policy retraining. Extensive experiments on three public benchmarks, 500 customized tasks, and 135 real-robot experiments demonstrate effective capability-aware routing and stable policy orchestration, yielding substantial improvements in zero-shot long-horizon planning and out-of-distribution robustness.

Graphical User Interface (GUI) agents powered by vision-language models hold promise for automating real-world mobile tasks. However, progress is limited by the lack of high-coverage, long-horizon interaction trajectories collected from element-rich and rapidly evolving apps. Existing pipelines often rely on costly human demonstrations or on-policy framework, which tends to over-sample common flows while missing rare transitions and complex multi-step procedures. To address this problem, we propose SEE, a two-stage data synthesis framework consisting of (i) an efficient exploration stage that builds an explicit UI transition graph over screens and elements, and (ii) a graph-based synthesis stage that composes diverse multi-step trajectories via planning and controlled sampling. This design yields reproducible and explainable data generation, while explicitly preventing spurious cycles and enabling long-horizon composition. Across multiple real-world apps, SEE produces trajectories with an average length of 14.8 steps while avoiding spurious loops, and agents fine-tuned on SEE achieve improved task success and generalization to unseen screens. We will publicly release our synthesis code and dataset.

Latent world models have emerged as a powerful planning paradigm by learning action-conditioned predictive dynamics and using them as internal simulators to imagine and evaluate candidate action sequences. However, as the planning horizon grows, performance becomes increasingly constrained by proposal quality: a fixed candidate budget must search an exponentially larger action space, making it difficult to expose the world model to high-quality candidate futures for evaluation. In this paper, we introduce a prior-conditioned planner that replaces random proposal initialization with structured guidance. At each planning stage, a goal-conditioned generator predicts the next reachable latent subgoal for a specified duration, which is then used to condition the generation of candidate action sequences. To capture semantic information across temporal scales, we use subgoals of varying durations as priors, balancing fine-grained local control with higher-level long-horizon progress. Then the frozen world model evaluates and refines these subgoal-conditioned proposals before execution. Experiments on PushT and OGBench Cube show that coupling latent subgoal decomposition with prior-conditioned action generation substantially improves long-horizon planning while preserving strong short-horizon performance. To be specific, when the target offset is $150$, it raises PushT success from $12.7\%$ to $64.7\%$ and OGBench Cube success from $26.7\%$ to $67.3\%$.

Coordinating payload transfers between subsystems is a critical challenge in lifelong Multi-Agent Pickup and Delivery (MAPD). We study systems where agents are confined to separate regions and must exchange payloads through shared handover stations. These stations, equipped with single docks and finite buffers, are inherently vulnerable to blocking and starvation. We formalize this problem as Multi-Subsystem MAPD with Buffer-limited Handover Stations (MS-MAPD-BHS). We then propose Handover-Aware Reservation and Routing (HARR), an online controller that couples per-subsystem planners. HARR uses a shared dock reservation calendar and a deterministic rolling-horizon projection of buffer occupancy to coordinate actions. A candidate route is accepted only if its dock interval is free and the resulting buffer occupancy projection remains within capacity. Under perfect execution, these checks ensure collision-free dock use and buffer-safe committed operations within the reservation horizon. In simulation, HARR achieves up to 77% higher throughput and 92% lower backlog than a fixed-dock ablation at moderate load, while also reducing planning time relative to a coupled station-aware Token Passing baseline. These results show that explicit interface coordination substantially improves stability in modular multi-subsystem transport.

This paper addresses key technical challenges in current large language model (LLM) agent applications, including long-horizon planning, sparse reward attribution, and dynamic environmental interaction, by designing and optimizing an intelligent agent workflow. The proposed architecture is based on the synthesis of core AI paradigms: Visual, Language, Generative, Graph, Multimodal, Reinforcement, and Agent Intelligence. Unlike conventional baseline models that rely on static prompting and lack robust perception-action loops, our approach introduces a Partially Observable Markov Decision Process (POMDP) routing mechanism. This mechanism is augmented with an internal, self-correcting reward model that evaluates decision trajectories before execution. By integrating multimodal inputs and advanced reinforcement learning principles (such as proximal policy optimization and value function approximation), the agent maintains long-term structural memory and dynamically adapts its reasoning pathways to mitigate error accumulation. Empirical experiments on the ALFWorld embodied simulation environment and the WebShop online navigation benchmark demonstrate a 24.5% absolute improvement in task success rate and trajectory efficiency over mainstream baselines like the standard ReAct framework. Comprehensive ablation studies confirm the significant contribution of the reward-driven critique module in suppressing hallucination rates. This research bridges theoretical foundations of reinforcement learning and graph-based memory with autonomous agent workflows. Ultimately, the resulting architecture offers a practical, scalable reference framework for developing artificial intelligence technologies in complex, multi-step autonomous systems. Code is available at https://github.com/01Amez/RLAW_Implementation.

Existing memory systems for long-horizon LLM agents often retrieve prior traces as passive context rather than converting them into executable capabilities. In this paper, we propose MSCE, a training-free Memory--Skill Co-Evolution framework that organizes agent experience into grounded step traces, reusable procedural policies, and declarative environmental cognition. MSCE crystallizes evidence-backed L2 policies with positive estimated gain into callable skills that retain evidence links, applicability boundaries, decision guidance, verification rules, and reliability estimates. It further introduces reflection-weighted value backfilling, which propagates sparse terminal feedback through dense local self-reflections to produce evidence-calibrated trace values for governing memory and skill evolution. Experiments on EvoAgentBench and LoCoMo demonstrate that MSCE significantly outperforms state-of-the-art skill-augmented and memory-driven agent baselines, exhibiting strong cross-domain transferability and lifelong-evolution capabilities.

Long-horizon AI agents are becoming increasingly capable, yet their interaction with users remains surprisingly thin. In most workflows, users give an initial instruction, receive only selective textual updates, and lose a clear sense of what the agent is doing or when to step in. This leaves a missing part in the current agent ecosystem: an always-on Jarvis-style mediator that keeps the agent continuously reachable to the user. Such a mediator should support real-time spoken interaction with the user, answer questions without interrupting the worker, proactively report progress or confusion, and inject user guidance back into the agent's execution when useful. In this work, we introduce JarvisBench, a benchmark for measuring the dual value of mediation in long-horizon agent workflows. JarvisBench contains two complementary tracks: an agent-collaboration track that measures whether mediation improves downstream task completion, and a user-interaction track that measures whether mediation makes ongoing execution more understandable, responsive, and accessible to users. We instantiate the benchmark with a modular reference Jarvis prototype and evaluate it on 34 text-only WildClaw tasks executed in OpenClaw. Preliminary results with GPT-5.5, Claude Opus 4.7, Gemini-based, and GPT-based worker agents suggest that Jarvis-style mediation can provide trace-grounded responses to user questions and improve task performance when sparse user guidance is injected at appropriate moments. The results also show that effectiveness depends strongly on the mediator's LLM brain, highlighting both the promise of this missing middle layer and the need for broader community effort. Demo page https://cchen1436.github.io/jarvis

Vision-Language-Action (VLA) policies offer strong general-purpose manipulation priors, but often fail on tight-tolerance, contact-rich assembly due to long-horizon credit assignment and subtask coupling: a state that is geometrically successful for the current skill can be brittle for downstream skills. We show this failure mode in residual reinforcement learning (RL) over a frozen VLA base policy: constant sparse success rewards improve each subtask in isolation yet yield little or no gain when skills are chained, because terminal state quality is uncontrolled. We propose Foresight Residual RL, which optimizes handoff quality by augmenting each subtask's sparse success reward with an offline-estimated foresight value -- the probability of future subtask success conditioned on the terminal state of the current subtask. Concretely, we (i) train a visual foresight predictor from images of terminal states of the base policy, labeled using downstream rollout statistics, and (ii) train residual policies via backward foresight induction, using the predictor output as a reward multiplier. On a three-phase wrench-based nut-tightening assembly task in Isaac Gym (grasp, move-insert, rotate), our method achieves 85.6% full-task success, outperforming standard subtask residual RL (54.5%) and VLA baselines, while leaving per-subtask success unchanged. These results highlight that improving long-horizon performance requires shaping which successful states are produced at each sub-task, not only whether success occurs.

Real-time collision avoidance for robotic manipulators requires fast reactions to unexpected obstacle motion and lookahead to avoid becoming trapped by near-future constraints. Full model predictive control can provide this foresight, but its online cost may grow quickly with horizon length, model fidelity, and the number of active geometric constraints. Conversely, horizon-free reactive methods are computationally efficient but can be short-sighted in dynamic clutter. We present a task-space receding-horizon controller that uses a short contact-consistent rollout to generate a terminal kinematic reference satisfying internal non-penetration constraints, then computes only the first input of a smooth minimum-acceleration transition toward that reference. Starting from a closed-loop inverse-kinematics regulation law, the rollout is performed with an iterative dynamics solver operating on inflated convex robot and obstacle geometries, so that robot-obstacle contacts, dynamic obstacle motion, and self-collisions can shape the terminal reference without requiring full constrained trajectory optimization. We analyze the contact-inactive closed loop and show local exponential task-space regulation under standard regularity assumptions. For contacts activated inside the rollout, we characterize the corresponding discrete updates and bound the effect of moving obstacles on regular operating sets. Simulations on a 40-DOF multi-chain system show that intermediate horizons balance anticipation, responsiveness, and computational cost. Hardware experiments on a 6-DOF platform demonstrate consistent sim-to-real behavior without accurate inertial parameter estimation, and comparisons against dynamic optimization fabrics and model predictive control (MPC) baselines show improved success rates in dynamic clutter while preserving solve times compatible with real-time execution in the tested regimes.

While LLM agents demonstrate strong reasoning abilities in compact and well-defined scenarios, they struggle to maintain robustness and effectiveness when faced with large-scale, diverse, and dynamic real-world environments that demand seamless tool integration. To address this gap, we introduce ToolVerse, a comprehensive framework that scales up agentic RL environments and enables agents to perform complex long-horizon reasoning in Tool-Integrated Reasoning (TIR) tasks. First, ToolVerse automatically builds the massive executable agent training environments from nearly 400 real-world Model Context Protocols (MCPs) that contain about 4500 tools. Second, we propose a task design strategy based on a tool dependency graph, utilizing Dynamic Unlocking Sampling Algorithm to generate long-horizon tasks, and produce GUST (Graph Unlocking Sampling Tasks) dataset. Third, to alleviate the credit assigment problem in long-horizon agentic RL, we propose a fine-grained Turn-Aware Relative Advantage algorithm. We conduct extensive Agentic RL training using ToolVerse and evaluate our framework on serveral agentic benchmarks. Experimental results demonstrate that our framework significantly strengthens LLMs' capabilities in long-horizon tool use, achieving a marked performance boost and showcasing robust reasoning within dynamic environments.

Developing autonomous hydraulic excavators is constrained by limited access to physical machines and the high cost of real-world experimentation. This paper proposes a simulation-to-real framework for learning a system-level digital surrogate using Long Short-Term Memory (LSTM) networks. Instead of modeling internal dynamics, the excavator is treated as an input-output operator, and the surrogate is trained to reproduce its closed-loop behavior under identical control inputs. The approach is first validated in a MuJoCo simulation environment and then transferred to a real excavator. To address measurement inconsistencies in real-world data, a consistency-aware state estimation method based on adaptive Kalman filtering is introduced. Experimental results demonstrate that the learned surrogate achieves high fidelity in both angular velocity and long-horizon trajectory reproduction under closed-loop autoregressive evaluation. These results confirm that the proposed model can serve as a drop-in surrogate for both simulation and physical systems, enabling scalable and efficient development of excavation automation algorithms.

This article explores the application of covariate balance diagnostics for detecting the presence of hidden confounding/model miss-specification in studies applying offline reinforcement learning (RL) to deriving optimal treatment recommendations. The results demonstrate that, either there is a high risk of bias within existing offline RL studies for treatment recommendations or, existing covariate balance metrics are not sufficient to assess such studies. Regardless, existing offline RL studies cannot be concluded as being statistically robust. The conclusions propose future research directions for obtaining more methodologically robust applications of offline RL to treatment recommendation problems.

Multi-turn agents solve complex tasks through extended sequences of tool interactions before producing a final answer, making credit assignment a fundamental challenge during post-training. Outcome rewards provide reliable supervision for short-horizon reasoning, but become sparse and high-variance as trajectories grow to tens or hundreds of tool calls. They can also be misleading: a failed rollout may contain many useful actions that move the agent closer to the goal, yet outcome-only training assigns them the same negative advantage as the eventual mistake. We propose TRACE (Turn-level Reward Assignment via Credit Estimation), a dense credit-assignment method for agentic reinforcement learning. TRACE represents rollouts as state transitions at tool-call boundaries, obtains gold-answer log-probabilities from a frozen reference model, transforms them into log-ratio state values, and derives per-action rewards as Temporal-Difference changes in those values. This requires no additional critic or process-label training, and its one-step log-ratio TD component telescopes across redundant tool calls. On long-horizon complex search, TRACE substantially improves base-model tool-use ability using pure RL, without a cold-start supervised fine-tuning stage, an agentic mid-training stage, or training on live-web data. On the closed-web BrowseComp-Plus benchmark, it raises Qwen3-4B from $7.2$ to $35.6$ and Qwen3-30B-A3B from $8.4$ to $42.6$. The learned search behavior also transfers to open-web benchmarks, and the learning curves show earlier improvement and faster convergence during RL training.