ICML 2026 Papers

Agentic systems solve complex tasks by coordinating multiple agents that iteratively reason, invoke tools, and exchange intermediate results. To improve robustness and solution quality, recent approaches deploy multiple agent teams running in parallel to explore diverse reasoning trajectories. However, parallel execution comes at a significant computational cost: when different teams independently reason about similar sub-problems or execute analogous steps, they repeatedly perform substantial overlapping computation. To address these limitations, in this paper, we propose Learning to Share (LTS), a learned shared-memory mechanism for parallel agentic frameworks that enables selective cross-team information reuse while controlling context growth. LTS introduces a global memory bank accessible to all teams and a lightweight controller that decides whether intermediate agent steps should be added to memory or not. The controller is trained using stepwise reinforcement learning with usage-aware credit assignment, allowing it to identify information that is globally useful across parallel executions. Experiments on the AssistantBench and GAIA benchmarks show that LTS significantly reduces overall runtime while matching or improving task performance compared to memory-free parallel baselines, demonstrating that learned memory admission is an effective strategy for improving the efficiency of parallel agentic systems.

Multimodal Retrieval-Augmented Generation (MRAG) has emerged as a key paradigm for grounding MLLMs with external knowledge. While query pre-processing (e.g., rewriting) is standard in text-based RAG, existing MRAG pipelines predominantly treat visual inputs as static and immutable, implicitly assuming they are noise-free. However, real-world visual queries are often ``imperfect''---suffering from geometric distortions, quality degradation, or semantic ambiguity---leading to catastrophic retrieval failures. To address this gap, we propose V-QPP-Bench, the first comprehensive benchmark dedicated to Visual Query Pre-processing (V-QPP). We formulate V-QPP as an agentic decision-making task where MLLMs must autonomously diagnose imperfections and deploy perceptual tools to refine queries. Our extensive evaluation across 46,700 imperfect queries and diverse MRAG paradigms reveals three critical insights: (1) Vulnerability---visual imperfections severely degrade both retrieval recall and end-to-end MRAG performance; (2) Restoration Potential \& Bottleneck---while oracle preprocessing recovers near-perfect performance, off-the-shelf MLLMs struggle with tool selection and parameter prediction without specialized training; and (3) Training Enhancement---supervised fine-tuning enables compact models to achieve comparable or superior performance to larger proprietary models, demonstrating the benchmark's value for developing robust MRAG systems The code is available at https://github.com/phycholosogy/VQQP_Bench

In recent years, LLM-based multi-agent systems (MAS) have advanced rapidly, using a router to decompose tasks and delegate subtasks to specialized agents. A natural way to expand capability is to scale up the agent pool by continually integrating new functional agents or tool interfaces, but naive expansion can trigger performance collapse when the router cold-starts on newly added, heterogeneous, and unreliable agents. We propose MonoScale, an expansion-aware update framework that proactively generates a small set of agent-conditioned familiarization tasks, harvests evidence from both successful and failed interactions, and distills it into auditable natural-language memory to guide future routing. We formalize sequential augmentation as a contextual bandit and perform trust-region memory updates, yielding a monotonic non-decreasing performance guarantee across onboarding rounds under a non-interfering expansion assumption. Experiments on GAIA and Humanity's Last Exam show stable gains as the agent pool grows, outperforming naive scale-up and strong-router fixed-pool baselines. Our code is available here.

Modern LLM serving systems must sustain high throughput while meeting strict latency SLOs across two distinct inference phases: compute-intensive prefill and memory-bound decode phases. Existing approaches either (1) aggregate both phases on shared GPUs, leading to interference between prefill and decode phases, which degrades Time-Between-Tokens (TBT); or (2) disaggregate the two phases across GPUs, improving latency but wasting resources through duplicated models and KV cache transfers. We present DuetServe, a unified LLM serving framework that achieves disaggregation-level isolation within a single GPU. DuetServe operates in aggregated mode by default and dynamically activates SM-level GPU spatial multiplexing when TBT degradation is predicted. Its key idea is to decouple prefill and decode execution only when needed through fine-grained, adaptive SM partitioning that provides phase isolation only when contention threatens latency service level objectives. DuetServe integrates (1) an attention-aware roofline model to forecast iteration latency, (2) a partitioning optimizer that selects the optimal SM split to maximize throughput under TBT constraints, and (3) an interruption-free execution engine that eliminates CPU–GPU synchronization overhead. Evaluations show that DuetServe improves total throughput by up to 1.3$\times$ while maintaining low generation latency compared to state-of-the-art frameworks.

Multi-turn, multi-agent LLM game evaluations exhibit substantial run-to-run variance. In long-horizon interactions, small deviations compound across turns and are amplified by multi-agent coupling, biasing win rate estimates and destabilizing comparative rankings across repeated tournaments. Prompt choice exacerbates this by inducing different effective policies and interaction dynamics. We address both instability and underperformance in games with **MEMO:** (**Me**mory-augmented **MO**del context optimization), a self-play framework that treats inference-time context as an optimizable, agentic object by coupling **retention** and **exploration**. Retention distills self-play trajectories into a persistent memory bank of structured insights that act as priors; exploration combines tournament-style prompt evolution under TrueSkill with prioritized replay over decisive states. Across five text-based games and a 2,000-game budget per task, MEMO raises mean win rate from 25.1% to 49.5% for GPT-4o-mini and from 20.9% to 45.0% for Qwen-2.5-7B, while sharply reducing run-to-run dispersion. These results suggest substantial headroom in multi-agent LLM game performance and robustness can be unlocked through memory, with MEMO achieving gains in negotiation and imperfect-information settings, while RL remains more effective in perfect-information settings. Project website: https://yunfeixie233.github.io/MEMO/.

Multimodal Large Language Models (MLLMs) excel at broad visual understanding but still struggle with fine-grained perception, where decisive evidence is small and easily overwhelmed by global context. Recent "Thinking-with-Images" methods alleviate this by iteratively zooming in and out regions of interest during inference, but incur high latency due to repeated tool calls and visual re-encoding. To address this, we propose Region-to-Image Distillation, which transforms zooming from an inference-time tool into a training-time primitive, thereby internalizing the benefits of agentic zooming into a single forward pass of an MLLM. In particular, we first zoom in to micro-cropped regions to let strong teacher models generate high-quality VQA data, and then distill this region-grounded supervision back to the full image. After training on such data, the smaller student model improves "single-glance" fine-grained perception without tool use. To rigorously evaluate this capability, we further present ZoomBench, a hybrid-annotated benchmark spanning six fine-grained perceptual dimensions, together with a dual-view protocol that quantifies the global-regional "zooming gap". Experiments show that our models achieve leading performance across multiple fine-grained perception benchmarks (Figure 1), and also improve general multimodal cognition on benchmarks such as visual reasoning and GUI agents. Our code is available at https://github.com/inclusionAI/Zooming-without-Zooming.

Advancing complex reasoning in large language models relies on high-quality, verifiable datasets, yet human annotation remains cost-prohibitive and difficult to scale. Current synthesis paradigms often face a recurring trade-off: maintaining structural validity typically restricts problem complexity, while relaxing constraints to increase difficulty frequently leads to inconsistent or unsolvable instances. To address this, we propose \textbf{Agentic Proposing}, a framework that models problem synthesis as a goal-driven sequential decision process where a specialized agent dynamically selects and composes modular reasoning skills. Through an iterative workflow of internal reflection and tool-use, we develop the \textbf{Agentic-Proposer-4B} using Multi-Granularity Policy Optimization (MGPO) to generate high-precision, verifiable training trajectories across mathematics, coding, and science. Empirical results demonstrate that downstream solvers trained on agent-synthesized data significantly outperform leading baselines and exhibit robust cross-domain generalization. Notably, a 30B solver trained on only 11,000 synthesized trajectories achieves a state-of-the-art 91.6\% accuracy on AIME25, rivaling frontier-scale proprietary models such as GPT-5 and proving that a small volume of high-quality synthetic signals can effectively substitute for massive human-curated datasets.

Scientific illustrations are essential for depicting conceptual designs, methodologies, and experimental workflows in research, playing a pivotal role in communicating complex academic insights. However, creating high-quality scientific illustrations remains a labor-intensive task for human scientists. While recent generative image models have advanced prompt-based editing, the synthesis of fully **editable** figures remains a fundamental challenge. Valid editability involves structured transformations of graphical elements, scales, attributes, and text, rather than simple pixel-level changes. Existing models generate raster outputs that do not support manual correction or layout adjustment, limiting their utility in scientific publishing, where editable vector figures are typically required for submission. To address this challenge, we introduce **LiveFigure**, an agentic framework driven by VLM agents that imitates the multi-step drawing workflow of human researchers. It first plans figure blueprints by drawing inspiration from high-quality references in previous works, then generates executable scripts that produce figures via the PowerPoint interface based on skills and experience, and finally refines the outputs with targeted visual diagnostics, producing fully vectorized, editable figures that meet publication standards. Extensive experiments demonstrate that LiveFigure generates inherently editable figures, achieving 80% publication-readiness in only 17 manual edits, far surpassing the 24% rate of the strongest baseline, NanoBanana. Human preference studies further validate this advantage, with LiveFigure securing a 60% win rate against NanoBanana. Our code is available at [https://github.com/tsinghua-fib-lab/LiveFigure.git](https://github.com/tsinghua-fib-lab/LiveFigure.git).

As LLM-based agents are deployed in increasingly complex real-world settings, existing benchmarks underrepresent key challenges such as enforcing global constraints, coordinating multi-tool reasoning, and adapting to evolving user behavior over long, multi-turn interactions. To bridge this gap, we introduce \textbf{TRIP-Bench}, a long-horizon benchmark grounded in realistic travel-planning scenarios. TRIP-Bench leverages real-world data, offers 18 curated tools and 40+ travel requirements, and supports automated evaluation. It includes splits of varying difficulty; the hard split emphasizes long and ambiguous interactions, style shifts, feasibility changes, and iterative version revision. Dialogues span up to 15 user turns, can involve 150+ tool calls, and may exceed 200k tokens of context. Experiments show that even advanced models achieve at most 50\% success on the easy split, with performance dropping below 10\% on hard subsets. We further propose GTPO, an online multi-turn reinforcement learning method with specialized reward normalization and reward differencing. Applied to Qwen2.5-32B-Instruct, GTPO improves constraint satisfaction and interaction robustness, outperforming Gemini-3-Pro in our evaluation. We expect TRIP-Bench to advance practical long-horizon interactive agents, and GTPO to provide an effective online RL recipe for robust long-horizon training.

LLM agents increasingly act on users’ personal information, yet existing privacy defenses remain limited in both design and adaptability. Most prior approaches rely on static or passive defenses, such as prompting and guarding. These paradigms are insufficient for supporting contextual, proactive privacy decisions in multi-step agent execution. We propose *Contextualized Defense Instructing (CDI)*, a new privacy defense paradigm in which an instructor model generates step-specific, context-aware privacy guidance during execution, proactively shaping actions rather than merely constraining or vetoing them. Crucially, CDI is paired with an experience-driven optimization framework that trains the instructor via reinforcement learning (RL), where we convert failure trajectories with privacy violations into learning environments. We formalize baseline defenses and CDI as distinct intervention points in a canonical agent loop, and compare their privacy–helpfulness trade-offs within a unified simulation framework. Empirically, CDI consistently achieves a better balance between privacy preservation (94.2\%) and helpfulness (80.6\%) than baselines, with superior robustness to adversarial conditions and generalization.

Autonomous image-editing agents powered by multimodal large language models (MLLMs) improve transparency and controllability by translating high-level instructions into tool-mediated edit sequences, but training such agents with reinforcement learning often relies on dense proxy rewards (e.g., incremental image-quality score gains) to compensate for sparse human feedback. When these proxies overvalue small local changes, the resulting optimization signal can be dominated by numerically measurable yet perceptually negligible edits, biasing policy gradients toward proxy artifacts rather than meaningful progress. We propose B-Spar, a reward-centric Reinforcement Learning framework for perceptually aligned image retouching under sparse feedback that combines prior-guided trajectory sampling to reduce inefficient exploration, Bayesian reward modeling to densify sparse binary feedback into a stable training signal, and anchor-regularized policy optimization to steer updates toward high-reward regions while preventing early mode collapse. Experiments on public benchmarks demonstrate that B-Spar improves perceptual quality and metric alignment with stable training and competitive inference efficiency over strong prompt-based and training-based baselines. Notably, it outperforms AIGC-based baselines by over 95\% in perceptual quality, achieving an improvement of approximately 33.5\% over the state-of-the-art.

MLLM-powered embodied agents deployed in real-world environments encounter physical hazards. However, existing approaches lack explicit mechanisms for identifying hazards and reasoning about action-conditioned risks, leading agents to either miss risky interactions or over-identify risks. To address this, we propose EMBGuard, the first MLLM-based safety guardrail for embodied agents designed to decouple physical risk reasoning from agent policy. By evaluating a (visual observation, action) pair, EMBGuard identifies hazardous configurations and provides natural language explanations of potential risks. Alongside EMBGuard, we contribute EMBHazard, a training dataset of 15.1K action-conditioned pairs, and EMBGuardTest, a benchmark of 329 manually curated real-world scenarios spanning seven physical risk categories. Through compositional variation of hazards and actions, we generate diverse risky and benign scenarios that agents may encounter during planning. Despite its compact size (2B, 4B), EMBGuard achieves performance competitive with proprietary MLLMs (e.g., GPT-5.1, Gemini-2.5-Pro) while significantly reducing the false-positive rates that hinder real-time deployment. We make the code, data, and models publicly available at https://github.com/dongwxxkchoi/EMBGuard.

Current repository agents encounter a reasoning disconnect due to fragmented representations, as existing methods rely on isolated API documentation or dependency graphs that lack semantic depth. We consider repository comprehension and generation to be inverse processes within a unified cycle: generation expands intent into implementation, while comprehension compresses implementation back into intent. To address this, we propose RPG-Encoder, a framework that generalizes the Repository Planning Graph (RPG) from a static generative blueprint into a unified, high-fidelity representation. RPG-Encoder closes the reasoning loop through three mechanisms: (1) Encoding raw code into the RPG that combines lifted semantic features with code dependencies; (2) Evolving the topology incrementally to decouple maintenance costs from repository scale, reducing overhead by 95.7%; and (3) Operating as a unified interface for structure-aware navigation. In evaluations, RPG-Encoder establishes state-of-the-art repository understanding on SWE-bench Verified with 93.7% Acc@5 and exceeds the best baseline by over 10% on SWE-bench Live. These results highlight our superior fine-grained localization accuracy in complex codebases. Furthermore, it achieves 98.5% reconstruction coverage on RepoCraft, confirming RPG's high-fidelity capacity to mirror the original codebase and closing the loop between intent and implementation. Our code and data are available at https://github.com/microsoft/RPG-ZeroRepo.

Manufacturable chip layouts must satisfy thousands of geometry-based design rules, and design rule checking (DRC) enforces them by running executable DRC scripts on layouts. Translating natural language rules into correct DRC scripts is labor-intensive and requires specialized expertise, motivating LLM agents for DRC script synthesis and debugging. However, existing benchmarks have small evaluation sets and often evaluate scripts by code similarity rather than execution correctness, and prior machine learning-based methods either ignore execution feedback or require labeled test layouts as agent's input. To this end, we introduce Rule2DRC, a large-scale benchmark for DRC script coding agents with 1,000 rule-to-script tasks and 13,921 evaluation chip layouts for execution-based scoring. Rule2DRC provides an evaluation pipeline that measures functional correctness via DRC execution outcomes without requiring evaluation layouts as input to the agent. We also propose SplitTester, a tester agent for program selection that uses execution feedback to generate discriminative test cases and separate previously indistinguishable candidate scripts, substantially improving Best-of-N selection performance in this domain. We release the code at https://github.com/snu-mllab/Rule2DRC.

Deploying Large Language Models to data-scarce programming domains poses significant challenges, particularly for kernel synthesis on emerging Domain-Specific Architectures where a ``Data Wall'' limits available training data. While models excel on data-rich platforms like CUDA, they suffer catastrophic performance drops on data-scarce ecosystems such as NPU programming. To overcome this cold-start barrier without expensive fine-tuning, we introduce EvoKernel, a self-evolving agentic framework that automates the lifecycle of kernel synthesis from initial drafting to continual refining. EvoKernel addresses this by formulating the synthesis process as a memory-based reinforcement learning task. Through a novel value-driven retrieval mechanism, it learns stage-specific Q-values that prioritize experiences based on their contribution to the current objective—whether bootstrapping a feasible draft or iteratively refining latency. Furthermore, by enabling cross-task memory sharing, the agent generalizes insights from simple to complex operators. By building an NPU variant of KernelBench and evaluating on it, EvoKernel improves frontier models' correctness from 11.0\% to 83.0\% and achieves a median speedup of 3.60$\times$ over initial drafts through iterative refinement. This demonstrates that value-driven experience accumulation allows general-purpose models to master the kernel synthesis task on niche hardware ecosystems.

Large Language Models (LLMs) have enabled automated heuristic design (AHD) for combinatorial optimization problems (COPs), but existing frameworks' reliance on fixed evolutionary rules and static prompt templates often leads to myopic heuristic generation, redundant evaluations, and limited reasoning about how new heuristics should be derived. We propose a novel multi-agent reasoning framework, referred to as Planning through World Model for Automated Heuristic Design via Self-Evolving LLMs (PathWise), which formulates heuristic generation as a sequential decision process over an entailment graph serving as a compact, stateful memory of the search trajectory. This approach allows the system to carry forward past decisions and reuse or avoid derivation information across generations. A policy agent plans evolutionary actions, a world model agent generates heuristic rollouts conditioned on those actions, and critic agents provide routed reflections summarizing lessons from prior steps, shifting LLM-based AHD from trial-and-error evolution toward state-aware planning through reasoning. Experiments across diverse COPs show that PathWise converges faster to better heuristics, generalizes across different LLM backbones, and scales to larger problem sizes.

Reward shaping can accelerate reinforcement learning, but in sparse-reward \emph{offline} multi-agent RL it is often brittle: dense intrinsic rewards may alter the underlying Markov game, while world-model guidance can amplify model bias. We find that shaping becomes reliable when it is (i) activated only after \emph{statistically validated} learning plateaus and (ii) constrained to \emph{potential-based} shaping, which preserves the task optimum. Motivated by this, we propose \textsc{SOLAR}, a simulate--evaluate--shape framework. A learned world model enables low-cost rollouts to test plateaus; once a plateau is detected, we inject shaping in the form $r+\gamma\Phi(s')-\Phi(s)$ with adaptively updated potentials; and we attenuate shaping using uncertainty-aware throttling in unreliable regions. We provide theoretical analysis on policy invariance and on the deviation of plateau decisions under model error, and establish stability for the resulting two-timescale adaptation. Experiments on sparse-reward offline MARL benchmarks show consistent gains in stability and final performance across dataset qualities.

Agentic Web is an emerging paradigm where autonomous agents help users use online information. As the paradigm develops, content providers are also deploying agents to manage their data and serve it through controlled interfaces. This shift moves information access from centralized retrieval to decentralized coordination. To study this setting, we introduce AgentWebBench, a benchmark that evaluates how well a user agent synthesizes answers by interacting with website-specific content agents. We evaluate four tasks that cover common web information needs, spanning ranked retrieval (web search, web recommendation) and open-ended synthesis (question answering, deep research). Across seven advanced LLMs and three coordination strategies, multi-agent coordination generally lags behind centralized retrieval as expected, because user agent cannot directly access the corpus, but the gap shrinks with model scale and can even outperform centralized retrieval on question answering. This benchmark also enables us to study properties of the emerging paradigm of the digital world. We find that decentralized access concentrates traffic toward a small set of websites, test time scaling improves both interaction reliability and task performance, and strong results require sufficient interactions guided by careful planning. Finally, our failure analysis suggests that user agents need better planning and answer synthesis, while content agents need more reliable retrieval and evidence quality. Code, data, and APIs are released on https://github.com/cxcscmu/AgentWebBench.

Recent progress in multi-turn reinforcement learning (RL) has significantly improved reasoning LLMs' performances on complex interactive tasks. Despite advances in stabilization techniques such as fine-grained credit assignment and trajectory filtering, instability remains pervasive and often leads to training collapse. We argue that this instability stems from inefficient exploration in multi-turn settings, where policies continue to generate low-information actions that neither reduce uncertainty nor advance task progress. To address this issue, we propose Token- and Turn-level Policy Optimization (T$^2$PO), an uncertainty-aware framework that explicitly controls exploration at fine-grained levels. At the token level, T$^2$PO monitors uncertainty dynamics and triggers a thinking intervention once the marginal uncertainty change falls below a threshold. At the turn level, T$^2$PO identifies interactions with negligible exploration progress and dynamically resamples such turns to avoid wasted rollouts. We evaluate T$^2$PO in diverse environments, including WebShop, ALFWorld, and Search QA, demonstrating substantial gains in training stability and performance improvements with better exploration efficiency. Code is available at https://github.com/WillDreamer/T2PO.