Large Reasoning Models

As LLMs gain stronger reasoning capabilities, their extended chain-of-thought introduces new degrees of complexity for defending against adversarial jailbreaks and prompt injection. We study consistency training, a family of fine-tuning objectives that enforce identical behavior on clean prompts and adversarial rewrites, and evaluate its two main variants, output-level (BCT) and activation-level (ACT), across five reasoning models. We formulate both methods as a prompt injection defense and find ACT to be competitive with other training-based defenses while requiring only self-supervised pairs of clean and wrapped prompts. Our experiments also generalize both techniques within the jailbreak setting, demonstrating that ACT remains more robust to adaptive attacks. We also provide mechanistic evidence that ACT's defense against jailbreaks is encoded as a roughly linear shift in activation space at the assistant-turn boundary. After ACT training, we can recover a single steering direction that controls refusal on reasoning models with minimal effect on benign inputs. We find that ACT remains robust even when the model's chain-of-thought is replaced with a compliant trace from the undefended base model, pivoting to refuse prefilled jailbreaks. Together, these results suggest that supervising internal representations is a surprisingly effective and interpretable approach to various forms of safety training in reasoning models.

Reinforcement learning has become a central paradigm for advancing reasoning in large language models, yet most existing methods still depend on stronger teacher models or heavily curated difficult datasets, limiting scalable capability improvement. In this paper, we introduce DenoiseRL, a reinforcement learning framework that substitutes external supervision with recovery-oriented optimization over failures from weak models. Instead of relying on stronger supervision or carefully engineered data, DenoiseRL learns directly from incorrect reasoning traces by converting them into opportunities for improvement, making training more scalable and less dependent on external resources. This yields a richer and more diverse learning signal, improving exploration efficiency from imperfect model behavior. As a result, DenoiseRL improves reasoning performance and overall training efficiency while reducing the need for expensive data curation or stronger teacher models. Empirically, DenoiseRL consistently outperforms strong on-policy RL baselines across competitive mathematical and general reasoning benchmarks and promotes stronger self-corrective behavior as training difficulty increases, highlighting an effective and scalable alternative pathway for improving reasoning in large language models.

We highlight a failure mode of large reasoning models on questions with insufficient information: models may recognize that a problem is under-specified, yet still continue reasoning and produce unsupported final answers instead of abstaining. We formalize this mismatch as the detection-to-abstention gap, where detected insufficiency fails to translate into final abstention. This gap is especially concerning in high-risk domains such as medical AI, where answers based on incomplete evidence can be more harmful than refusal. To close this gap, we propose Judge-Then-Solve (JTS), a trajectory-level reasoning-control framework that trains models to make an explicit answerability commitment before solution generation. Rather than treating abstention as a final-answer style, JTS casts it as a control decision: the model either proceeds to solve or terminates early based on its answerability judgment. We instantiate this policy through supervised warm-up and missing-premise reinforcement learning with consistency and length-shaping rewards. Experiments on dense and MoE reasoning models show that JTS substantially improves reliable abstention across datasets and pushes Abstention@Detection (A@D) to near-saturation, indicating that models not only detect missing information but also act on that detection. By terminating unanswerable trajectories immediately after the answerability judgment, JTS reduces unnecessary reasoning and improves inference efficiency when continued deliberation would amplify unsupported assumptions. We also observe that missing-premise training can alter reasoning behavior on difficult but answerable problems, reducing unproductive self-reflection. These results suggest that abstention under insufficient information is a key form of reasoning control for deploying reasoning models safely and efficiently.

Multimodal Large Reasoning Models introduce the reasoning paradigm, demonstrating strong capabilities on complex vision-language tasks. However, they still suffer from severe hallucinations. Existing training-based methods typically mitigate hallucinations through response-level direct preference optimization (DPO), where the Chain-of-Thought (CoT) and the final answer are treated as a monolithic output and optimized jointly. We reveal that this formulation performs similarly to answer-only optimization, suggesting that it primarily learns answer-level preference, while leaving CoT-level supervision insufficiently exploited. To address this issue, we explicitly formulate a CoT-oriented preference term and derive Reasoning-Conditioned Direct Preference Optimization (RC-DPO), which models the CoT as a condition for answer generation and contrasts the preference for the same preferred answer under different CoT conditions, promoting answer-supportive reasoning chain alignment. To further improve optimization, we introduce a reasoning-enhanced preference data generation strategy that employs Monte Carlo Tree Search to discover visually grounded and logically consistent CoTs as positive samples, and attention-guided CoT token pruning to construct negative ones. Extensive experiments across various models and benchmarks show that RC-DPO effectively mitigates hallucinations and improves the reliability of the multimodal reasoning process.

Large Reasoning Models are typically trained via reinforcement learning from verifiable rewards (RLVR). However, existing approaches adopt fixed weights for positive and negative samples, and the conclusions hardly generalize to open-ended question answering (QA). In this paper, we systematically investigate the roles of positive and negative samples in reinforcement learning for open-ended QA. We propose a reward-mean-based strategy for distinguishing positive from negative samples, and observe that negative samples predominantly govern response diversity and the performance upper bound, whereas positive samples primarily determine response quality and convergence stability. Building on these observations, we propose EAPO, an Entropy-driven Adaptive Policy Optimization method that adaptively computes the weighting coefficients of positive samples based on the ratio of the current policy entropy to the initial entropy. During the entropy-decreasing phase, the weight assigned to positive samples is reduced to preserve exploration, whereas during the entropy-increasing phase it is amplified to reinforce stability, thereby mitigating entropy collapse. Experiments on two publicly available open-ended medical QA datasets demonstrate that EAPO consistently and substantially outperforms fixed-weight baselines in both response diversity and stability.

User modeling aims to use language models (LMs) to mimic an individual's behavior from a corpus of past context-action pairs (e.g., conversation turns), enabling the simulation of users in settings like behavioral science, human-AI collaboration, and market research. Recent approaches augment these corpora with synthesized reasoning traces, typically generated by conditioning on both context and action. However, such conditioning constitutes post-hoc rationalization rather than reasoning: the trace is guaranteed to justify the action, but may not encode the underlying latent causal decision paths. We propose Recon, which uses action reconstruction to score reasoning traces by their predictive power: given a context and candidate reasoning, a reconstruction model predicts the action, and reconstruction fidelity determines reasoning quality. Across four domains, Recon achieves a 54.7% win rate over Backward Synthesis, a standard post-hoc rationalization baseline. Further, we find that training a reasoning synthesis model with rewards derived from Recon improves downstream user modeling performance, achieving a win rate of up to 70.0% over baselines. We further show that Recon-synthesized reasoning transfers across models, and improves user modeling beyond the reconstruction model. Our work demonstrates that post-hoc rationalization is insufficient for reasoning synthesis, and that useful and interpretable reasoning should naturally elicit the action from the context.

Large reasoning models (LRMs) generate chain-of-thought (CoT) traces before producing final outputs, introducing a dynamic internal state that may complicate control mechanisms such as refusal. Unlike instruction-tuned LLMs, where refusal is mediated by a single directional subspace, refusal in large reasoning models (LRMs) additionally depends on the CoT. In DeepSeek-R1-Distill-LLaMA-8B, activation steering reverses refusal in only 39% of cases when the CoT is kept fixed, but removing the CoT entirely increases this to 70%, indicating that the CoT actively reinforces refusal. In a two-stage intervention where the model regenerates its CoT under activation steering, refusal is reversed in 94% of cases, while the resulting CoT alone retains 48% of this effect even after steering is removed. This suggests that the CoT can carry and reconstruct the compliance signal independently. These findings indicate that refusal in LRMs is jointly encoded in residual stream activations and CoT. This joint activation makes LRM more robust against activation-level interventions alone, but exposes CoT to a possible alternative surface attack.

Post-training has significantly enhanced the reasoning capability of Large Reasoning Models (LRMs), especially with Reinforcement Learning (RL) like Group Relative Policy Optimization (GRPO). However, GRPO-style RL methods in multi-domain settings often fail to achieve consistent improvements across all domains due to inherent interference in policy optimization. Prior studies on multi-domain RL primarily focus on alleviating cross-domain interference, while often neglecting the pivotal role of knowledge sharing, which we argue is the key to transforming cross-domain interactions from harmful competition into beneficial transfer. To address this limitation, we propose Multi-domain Contrastive Policy Optimization (MCPO), which analyzes the structural relationships among rollouts and promotes cross-domain knowledge sharing and in-domain knowledge consolidation in a contrastive manner. Specifically, for a given prompt, MCPO identifies transferable reasoning trajectories from other domains as positive examples, while treating incorrect rollouts as negative ones. It then encourages consistent representations for positive pairs and pushes negative pairs apart, thereby facilitating knowledge transfer and reducing interference. Moreover, MCPO aligns intra-domain correct rollouts to build a consolidated representation space. In this way, MCPO contrastively learns a harmonious representation space that can accommodate diverse multi-domain knowledge. Empirical results show that MCPO improves the reasoning capabilities of LRMs across multiple domains and even outperforms single-domain training in some cases. Code is available at https://github.com/Maricalce/MCPO.

Self-improvement training enables the large reasoning models (LRMs) to improve themselves by self-generating reasoning trajectories as training data without external supervision. However, we find that this method often falls short in complex reasoning tasks and even leads to model collapse. Through a series of preliminary analyses, we reveal two problems: (1) data imbalance, where most training samples are simple, but the challenging yet crucial samples are scarce; (2) overthinking, where many undesired samples with redundant reasoning steps are used for self-training. To this end, we propose HSIR, which effectively Harnesses Self-Improvement in large Reasoning models via two simple-yet-effective approaches. Specifically, HSIR introduces a verify-then-exit sampling strategy to mitigate data imbalance by efficiently collecting more accurate solutions for difficult queries, and designs an Intrinsic Diversity score to quantify overthinking and filter out the undesired solutions. We apply HSIR to various post-training paradigms, among which we further propose H-GRPO, an enhanced GRPO algorithm that leverages the intrinsic diversity as an external reward to encourage concise and diverse reasoning via reinforcement learning. Extensive results show that HSIR not only effectively enhances the reasoning performance, i.e., bringing up to +10.9% average performance gains, but also significantly improves the reasoning efficiency by reducing up to 42.4% relative inference overhead.

Large Reasoning Models (LRMs) have demonstrated remarkable capabilities in reasoning and generation tasks and are increasingly deployed in real-world applications. However, their explicit chain-of-thought (CoT) mechanism introduces new security risks, making them particularly vulnerable to jailbreak attacks. Existing approaches often rely on static CoT templates to elicit harmful outputs, but such fixed designs suffer from limited diversity, adaptability, and effectiveness. To overcome these limitations, we propose an adaptive evolutionary CoT jailbreak framework, called AE-CoT. Specifically, the method first rewrites harmful goals into mild prompts with teacher role-play and decomposes them into semantically coherent reasoning fragments to construct a pool of CoT jailbreak candidates. Then, within a structured representation space, we perform multi-generation evolutionary search, where candidate diversity is expanded through fragment-level crossover and a mutation strategy with an adaptive mutation-rate control mechanism. An independent scoring model provides graded harmfulness evaluations, and high-scoring candidates are further enhanced with a harmful CoT template to induce more destructive generations. Extensive experiments across multiple models and datasets demonstrate the effectiveness of the proposed AE-CoT, consistently outperforming state-of-the-art jailbreak methods.

Can post-trained large language models (LLMs) further improve themselves using only unlabeled prompts, without external teachers or feedback from tools? We study this setting starting only from unlabeled seed questions with no ground-truth solutions, across three reasoning domains: math, science, and coding. We propose Self-Verified Distillation, a simple post-training refinement algorithm in which the model generates candidate solutions to these seed questions, filters them using prompt-based self-verification, and trains on the resulting self-curated dataset. Inspired by the UQ benchmark's use of multiple validators to screen candidate answers to hard unsolved questions, we adapt this validation-based filtering idea to self-training: the model filters its own generated solutions through a three-stage cascade of cycle-consistency, factuality, and correctness checks, accepting a solution only if it passes all stages with unanimous judge votes. We find that sampling more candidate generations and using a larger verification budget during training data construction produces higher-quality self-curated data and, in turn, better reasoning models. We then train Qwen3 models at multiple scales with Self-Verified Distillation and obtain gains across all three domains. For Qwen3-4B, our method improves aggregate held-out pass@1 by +16.7 points in math (AIME26 and HMMT), +11.1 points in science (GPQA Diamond and HLE), and +8.3 points in coding (LCBv5 and LCBv6), with gains also extending to 0.6B and 8B models. Compared to our test-time-only baseline (UQ-TTC), which improves performance by spending extra compute at inference time, Self-Verified Distillation achieves better performance in most settings while requiring only a single inference call at test time.

Mathematical reasoning is essential for problem-solving in education, science, and industry, serving as a crucial benchmark for evaluating artificial intelligence systems. As Large Language Models (LLMs) improve their reasoning capabilities, understanding how well they perform mathematical reasoning has become increasingly important. This survey synthesizes recent advancements in mathematical reasoning with LLMs through a structured analysis of datasets, architectures, training strategies, and evaluation protocols. Our systematic review encompasses approximately 120 peer-reviewed studies and preprints, examining the evolution of this research area and providing a unified analytical framework to understand current progress and limitations. Our study particularly introduces a unified taxonomy of mathematical datasets, distinguishing between pretraining corpora, supervised fine-tuning resources, and evaluation benchmarks across varying levels of reasoning complexity. A systematic analysis of reasoning architectures and training strategies, including tool integration, verifier-guided reasoning, and parameter-efficient adaptation, is presented to assess their effects on reasoning robustness and generalization. Moreover, a comparative evaluation of existing metrics highlights the gap between final-answer accuracy and process-level reasoning verification. By synthesizing insights across these areas, our analysis identifies recurring failure modes, such as reasoning faithfulness issues, benchmark biases, and generalization limitations, and outlines key research directions toward improving symbolic grounding, evaluation reliability, and the development of more robust and trustworthy LLM-based reasoning systems.

Large Language Model (LLM) agents increasingly act inside real workspaces, where tools and skills determine whether model reasoning becomes reliable action. Existing skills remain largely informal: Markdown skills and instruction packs encode procedures as long natural-language documents, while function calling, Model Context Protocol (MCP) servers, and framework tools structure individual actions but usually leave workflow state, policy enforcement, and completion discipline outside the skill itself. We introduce Formal Skill, a runtime-native abstraction that represents reusable capability with JSON metadata and action schemas, reliable Python executors, hook-governed control logic, Formal Skill routing, and skill-local runtime state. By moving reusable procedure from repeated prompt text into executable state machines and hook policies, Formal Skill gives agents a token-efficient and enforceable control surface. We implement the abstraction in FairyClaw, an open-source event-driven runtime for executable, observable, and composable Formal Skills. On Harness-Bench, FairyClaw obtains highly competitive average scores while using substantially fewer tokens, with especially strong results on tasks that expose the role of Formal Skill.

Large Reasoning Models (LRMs) have demonstrated remarkable capabilities in solving complex problems by generating structured, step-by-step reasoning content. However, exposing a model's internal reasoning process introduces additional safety risks; for example, recent studies show that LRMs are more vulnerable to jailbreak attacks than standard LLMs. In this paper, we investigate jailbreak attacks on LRMs and reveal that the attack success rate (ASR) is closely correlated with LRMs' attention patterns. Specifically, successful jailbreaks tend to assign lower attention to harmful tokens in the input prompt, while allocating higher attention to those tokens in the reasoning content. Motivated by this finding, we propose a novel jailbreak method for LRMs that leverages reinforcement learning (RL) to enhance attack effectiveness, explicitly incorporating attention signals into the reward function design. In addition, we introduce diverse persuasion strategies to enrich the RL action space, which consistently improves the ASR. Extensive experiments on five open-source and closed-source LRMs across three benchmarks demonstrate that our method achieves substantially higher ASR, outperforming existing approaches in terms of effectiveness, efficiency, and transferability.

Large Reasoning Models (LRMs) introduce new opportunities for safety monitoring through their Chain of Thought (CoT) reasoning. However, CoT is not always faithful to the model's final output, undermining its reliability as a monitoring tool. To address this, we investigate the hidden representations of LRMs to determine whether future behavior can be predicted from prompt and CoT representations. By evaluating a probe at each generated token, we construct a probe trajectory, the continuous evolution of a concept's probability across the reasoning process. We find that future model behavior is more distinguishable when examined over the full trajectory than from a single static prediction. To characterize these temporal dynamics, we extract signal-processing features that capture volatility, trend, and steady-state behavior, significantly improving the separation of future model states. We also present two methodological insights. First, template-based training data achieves near-parity with dynamically generated model responses, eliminating the need for a costly initial inference and labeling. Second, the choice of pooling operation is critical: average-pooling and last-token methods collapse to near-random performance, while max-pooling achieves up to 95% AUROC and yields stable probe trajectories. Using four datasets and four reasoning models across the domains of safety and mathematics, we demonstrate that trajectory features encode task-specific dynamics that improve outcome separability. These findings establish probe trajectories as a complementary framework for monitoring LRM behavior. Warning: This article contains potentially harmful content.

Code reasoning refers to the task of predicting the output of a program given its source code and specific inputs. It can measure the reasoning capability of large language models (LLMs) and also benefit downstream tasks such as code generation and mathematical reasoning. Existing work has verified the effectiveness of reinforcement learning on the task. However, these methods design rewards solely based on final outputs or coarse-grained signals, and neglect the inherent consistency of the stepwise reasoning process in the task. Therefore, these methods often result in sparse reward or reward hacking, which limits the full play of enhanced learning capabilities. To alleviate these issues, we propose CodeThinker, a consistency-driven reinforcement learning framework for code reasoning. Specifically, CodeThinker has three key components: (1) a stepwise reasoning-aware model training module, which utilizes a consistency tracing paradigm as a template to synthesize training data that captures the stepwise reasoning process; (2) a dynamic beam sampling strategy, which aims to improve the quality of sampled outputs under a fixed sampling budget; and (3) a consistency reward mechanism that can effectively alleviate reward hacking. Experiments on three popular benchmarks show that CodeThinker achieves state-of-the-art performance across multiple LLMs. For instance, it outperforms the strongest baseline by 4.3% in accuracy when deployed on Qwen2.5-Coder-7B-Instruct. We also validate the effectiveness of CodeThinker on downstream tasks. Results show that, without additional training, CodeThinker obtains average accuracy gains of 5.33 and 3.11 percentage points on mathematical reasoning and code reasoning tasks covering 17 programming languages, respectively.

The advancement of Large Reasoning Models (LRMs) has catalyzed a paradigm shift from reactive fast thinking'' text generation to systematic, step-by-step slow thinking'' reasoning, unlocking state-of-the-art performance in complex mathematical and logical tasks. However, the field faces \textit{the fundamental gap between token-level behavioral analysis and internal reasoning mechanisms, and the instability of reinforcement learning (RL) for reasoning optimization relying on costly external verifiers}. We identify and formally define \textbf{Entropy-Gradient Inversion}, a robust negative correlation between token entropy and logit gradients that acts as a definitive geometric fingerprint for LRM reasoning capability. Building on this, we propose \textbf{Correlation-Regularized Group Policy Optimization (CorR-PO)}, which embeds this inversion signature into RL reward regularization. Extensive experiments on various reasoning benchmarks across multiple model scales show CorR-PO consistently outperforms state-of-the-art baselines, confirming that stronger inversion directly correlates with superior reasoning performance.

Large Reasoning Models (LRMs) achieve strong performance by generating long chains of thought (CoT), but often overthink, continuing to reason after a solution has already stabilized and thereby wasting tokens and increasing latency. Existing inference-time early-exit methods rely primarily on answer-level signals, such as confidence or trial-answer consistency, to decide when to stop. However, these signals mainly reflect answer readiness rather than reasoning convergence: they may trigger before the model has finished exploring or self-correcting, causing premature exits that can degrade final-answer accuracy and leave the retained reasoning chain semantically incomplete. We identify reasoning-level semantic redundancy as a complementary signal for semantic-preserving early exit: when successive steps no longer add novel progress and instead revisit established conclusions, the reasoning trajectory has likely converged. Building on this insight, we propose PUMA, a plug-and-play framework that combines a lightweight Redundancy Detector with answer-level verification. The detector flags semantically redundant candidate exits, while verification confirms whether stopping is safe, allowing PUMA to remove redundant continuation while preserving both answer accuracy and a coherent reasoning prefix. Across five LRMs and five challenging reasoning benchmarks, PUMA achieves 26.2% average token reduction while preserving accuracy and retained CoT quality. Additional experiments on code generation, zero-shot vision-language reasoning, and learned stopping-policy internalization further demonstrate that reasoning-level redundancy is a robust, transferable, and learnable signal for efficient reasoning. Our code is available at \url{https://github.com/giovanni-vaccarino/PUMA}.

GRPO-style RLVR trains reasoning models from multiple on-policy attempts per prompt, but typically uses these attempts only through terminal rewards. We show that a mixed group contains a richer process signal: a correct completion is a self-generated witness of how the current policy can solve the problem, while a wrong completion provides on-policy prefixes where the policy needs correction. We introduce \emph{Self-Supervised On-Policy Distillation} (SSOPD), which distills a teacher distribution conditioned on the shortest correct completion into prefixes of the longest wrong completion. This converts intra-group correct--wrong contrast into dense process supervision without external solution traces. A stopping-time view motivates the shortest-correct / longest-wrong rule as a finite-group approximation to editing persistent failures toward fast-success actions, and a prompt-level frontier weight concentrates the auxiliary loss where correct and wrong branches coexist. Across AIME 2024, AIME 2025, and HMMT 2025, SSOPD improves over GRPO in all nine model-benchmark settings. On Qwen3-8B, it reaches a macro Avg@12 of 65.6, outperforming GRPO by 1.6 points and the solution-conditioned OPSD baseline by 0.8 points. Code will be released at https://github.com/tzq1999/SSOPD.

Recent progress in large language models has led to the emergence of reasoning models, which have shown strong performance on complex tasks through specialized fine-tuning procedures. While these methods reliably improve pass@1 accuracy, prior works have observed that they show a coverage shrinkage behavior, where pass@k degrades relative to the base model. In this paper, we investigate the cause of reasoning shrinkage under SFT-based post-training. We hypothesize that this behavior is driven by properties of the fine-tuning data, specifically related to decision points or "forks in the road" scenarios where model encounters indecipherable patterns with multiple valid reasoning paths. To test this hypothesis, we design controlled case studies that simulate such decision-point settings, spanning indecipherable nodes in graph branching, and reasoning modes. By tracking post-training dynamics in these settings, we find that the shrinkage phenomenon is tightly correlated with the prevalence of decision-point scenarios in the training data. We also demonstrate that this shrinkage behavior can be partially mitigated through targeted data synthesis design of decision-points and a more systematic diversity-encouraging decoding mechanism. Our findings identify data-centric factors as a key driver of shrinkage in reasoning models and highlight diversity-aware designs as an effective lever for controlling it. (Data and code for reproducing our experiments are available at https://github.com/psunlpgroup/reasoning_forks )

Large Reasoning Models (LRMs) generate chain-of-thought traces whose length tracks human reaction times across cognitive tasks, but recent debate questions whether this alignment reflects genuine computational structure or surface verbosity. We test whether the alignment varies with inference-time reasoning effort. Across GPT-OSS-20B and GPT-OSS-120B, three effort levels, and six reasoning tasks, within-task and cross-task alignment remain invariant: Bayes Factors lean toward the null, and mean alignment is numerically near-identical across conditions. A manipulation check reveals that the effort parameter sets an upper budget on generation rather than driving real-time allocation, suggesting that the allocation policy is crystallized at training time. Arithmetic complexity contrasts further show that token allocation tracks fine-grained, format-dependent human difficulty patterns, with model scale improving the match. Cognitive cost alignment between LRMs and humans appears to be a training-time achievement, robust to inference-time perturbations, supporting a compiled rather than online account of LRM problem-solving.

Large reasoning models (LRMs) substantially outperform their base LLM counterparts on challenging reasoning benchmarks, yet it remains poorly understood where base models go wrong during token-by-token generation and how to narrow this gap efficiently. We study the base-reasoning gap through quantifying token-level distributional disagreement between a base model and a stronger reasoning model using likelihood-based divergences. Across benchmarks, we find that the reasoning advantage is highly sparse and concentrates on a small set of early, planning-related decision tokens. For instance, on Qwen3-0.6B, only ~8% of generated tokens account for the salient disagreement, and these tokens concentrate early in the response, are strongly enriched in planning-related decisions (17x), and coincide with high base-model uncertainty -- suggesting that base models fail mainly at early planning points that steer the subsequent reasoning trajectory. Building on these findings, we propose disagreement-guided token intervention, a simple inference-time delegation scheme that performs a one-token takeover by the reasoning model only at high-disagreement positions and immediately switches back to the base model. With a small intervention budget, this sparse delegation substantially recovers and can even surpass the performance of a same-size reasoning model on challenging reasoning tasks. Code is available at https://github.com/AlphaLab-USTC/RRTokenIntervention.

Reasoning-trained language models often spend more tokens on harder problems, but longer chains of thought do not show whether a model is merely computing for more steps or following a different internal trajectory. We study this distinction through hidden-state trajectories during chain-of-thought generation across competitive programming, mathematics, and Boolean satisfiability. Raw trajectory geometry is strongly shaped by generation length: longer generations mechanically alter path statistics, so difficulty-dependent comparisons are misleading without adjustment. After residualizing trajectory statistics on length, difficulty remains systematically coupled to corrected trajectory geometry across all domains studied. The clearest reasoning-specific separation appears in the code domain, where harder problems show more direct corrected trajectories and less heterogeneous local curvature in reasoning-trained models than in matched instruction-tuned baselines. Corrected difficulty-geometry coupling is weaker, but still present, in mathematics and Boolean satisfiability. Prompt-stage linear probes do not mirror the code-domain separation, and behavioral annotations show that stronger corrected coupling co-occurs with strategy shifts and uncertainty monitoring. Together, these findings establish length correction as a prerequisite for generation-time trajectory analysis and show that reasoning training can be associated with distinct corrected trajectory geometry, with the strength of the effect depending on the domain.

Reactive synthesis, the problem of automatically constructing a hardware circuit from a logical specification, is a long-standing challenge in formal verification. It is elusive for two reasons: It is algorithmically hard, and writing formal specifications by hand is notoriously difficult. In this paper, we tackle both sides of the problem. For the algorithmic side, we present a neuro-symbolic approach to reactive synthesis that couples large reasoning models with model checkers to iteratively repair a synthesized Verilog implementation via sound symbolic feedback. Our approach solves more benchmarks than the best dedicated tools in the annual synthesis competition and extends to constructing parameterized systems, a problem known to be undecidable. On the specification side, we introduce an autoformalization step that shifts the specification task from temporal logic to natural language by introducing a hand-authored dataset of natural-language specifications for evaluation. We demonstrate performance comparable to that of starting from formal specifications, establishing natural synthesis as a viable end-to-end workflow.

KV-cache compression at small budgets is a crowded design space spanning cache representation, head-wise routing, compression cadence, decoding behavior, and within-budget scoring. We study seven mechanisms across these five families under matched mean cache on long-form mathematical reasoning (MATH-500~\cite{hendrycks2021math}) with two distilled-reasoning models (Qwen-7B and Llama-8B variants of DeepSeek-R1-Distill~\cite{deepseek2025r1}) at budgets $b \in \{64, 128\}$. All seven were rejected. We then propose $α$, a one-function modification to the TriAttention~\cite{mao2026triattention} retention scorer that replaces argmax-top-$k$ with greedy facility-location-inspired selection under a V-space redundancy penalty controlled by a single weight $λ$. A pre-registered protocol tunes $λ$ on a frozen development split and confirms on a disjoint held-out split; with $λ= 0.5$, $α$ clears Bonferroni on two of the four (model, budget) cells (Qwen $b{=}128$ and Llama $b{=}64$), no cell is significantly negative, and the pre-registered Branch~A triggers. The finding is asymmetric: a minimal scoring modification beat heavier structural redesigns in this regime, and the combined matched-memory, sympy-graded, held-out confirmation protocol is the evidence standard that made the asymmetry visible.

Can a committee of weak reasoning-model calls reach the performance of much stronger models? We study verifier-backed committee search as inference-time boosting for reasoning language models. The mechanism is not simply that ``more agents help'': samples expose latent correct solutions, while critics and comparators must recover them without access to the hidden verifier. We formalize this view by separating proposal coverage, local identifiability, progress, and diversity. We prove that coverage can be amplified by repeated sampling, but cannot by itself create useful critics or comparators; reliable amplification requires an additional local soundness signal, such as execution, proof checking, type checking, tests, or constraint solving. We give rank-based bounds showing when local selection errors compose into reliable trajectories, and characterize the proposer-side ceiling: oracle best-of-$k$ converges only to the mass of task slices on which the proposal system assigns nonzero useful probability. Empirically, on SWE-bench Verified, a single \texttt{GPT-5.4 nano} proposal solves $67.0\%$ of tasks. Using the same nano model, our critic--comparator orchestration reaches $76.4\%$ with $k=8$ proposals, matching the standalone performance of \texttt{Gemini 3 Pro} and \texttt{Claude Opus 4.5} Thinking and approaching the $79.0\%$ oracle best-of-$8$ upper bound. Thus, many correct patches are already present in weak-model proposal pools; the main challenge is selecting them. The remaining failures are mostly proposal-coverage failures, indicating shared blind spots that stronger selection alone cannot close.

Large Reasoning Models (LRMs) embedded in agentic frameworks have transformed information retrieval from static, long context question answering into open-ended exploration. Yet real world use requires models to discover and synthesize "long-tail" facts from dispersed sources, a capability that remains under-evaluated. We introduce PolitNuggets, a multilingual benchmark for agentic information synthesis via constructing political biographies for 400 global elites, covering over 10000 political facts. We standardize evaluation with an optimized multi agent system and propose FactNet, an evidence conditional protocol that scores discovery, fine-grained accuracy, and efficiency. Across models and settings, we find that current systems often struggle with fine-grained details, and vary substantially in efficiency. Finally, using benchmark diagnostics, we relate agent performance to underlying model capabilities, highlighting the importance of short-context extraction, multilingual robustness, and reliable tool use.

Large Reasoning Models (LRMs) are increasingly integrated into systems requiring reliable multi-step inference, yet this growing dependence exposes new vulnerabilities related to computational availability. In particular, LRMs exhibit a tendency to "overthink", producing excessively long and redundant reasoning traces, when confronted with incomplete or logically inconsistent inputs. This behavior significantly increases inference latency and energy consumption, forming a potential vector for denial-of-service (DoS) style resource exhaustion. In this work, we investigate this attack surface and propose an automated black-box framework that induces overthinking in LRMs by systematically perturbing the logical structure of input problems. Our method employs a hierarchical genetic algorithm (HGA) operating on structured problem decompositions, and optimizes a composite fitness function designed to maximize both response length and reflective overthinking markers. Across four state-of-the-art reasoning models, the proposed method substantially amplifies output length, achieving up to a 26.1x increase on the MATH benchmark and consistently outperforming benign and manually crafted missing-premise baselines. We further demonstrate strong transferability, showing that adversarial inputs evolved using a small proxy model retain high effectiveness against large commercial LRMs. These findings highlight overthinking as a shared and exploitable vulnerability in modern reasoning systems, underscoring the need for more robust defenses.

Recent multimodal large language models (MLLMs) have shown strong chain-of-thought (CoT) reasoning ability on vision-language tasks, but their direct deployment in real-world systems is often limited by latency and resource constraints. In practice, smaller MLLMs are preferred for online serving, yet their reasoning performance is bottlenecked by the lack of large-scale, high-quality multimodal CoT supervision. In this paper, we present OmniThoughtVis, a scalable data curation and distillation pipeline for transferring multimodal reasoning capabilities from high-capacity teacher models to smaller, deployment-oriented MLLMs. Starting from a diverse open-source seed pool, our pipeline generates structured CoT traces and performs joint annotation of reasoning difficulty, answer quality, and semantic task tags. To maintain data quality at scale, we combine rule-based filtering, difficulty-aware selection, and tag-based diversity sampling, resulting in a curated corpus of 1.8M samples that supports controllable subset construction for downstream training. We use OmniThoughtVis to distill Qwen3-VL models from 2B to 8B parameters and evaluate them on nine multimodal reasoning benchmarks. The resulting distilled models show consistent gains across model scales, including improvements of up to +16.8 points on MathVerse and +5.6 points on MMMU-Pro for the 4B model. Notably, the distilled 4B model matches or surpasses the undistilled 8B baseline on several tasks, highlighting the practical value of scalable reasoning distillation for deployment-oriented MLLMs.

Deep learning has become the dominant approach for creating high capacity, scalable models across diverse data modalities. However, because these models rely on a large number of learned parameters, tightly couple feature extraction with task objectives, and often lack explicit reasoning mechanisms, it is difficult for humans to understand how they arrive at their predictions. Understanding what representations emerge and why they arise from the training data remains an open challenge. We introduce Deep Arguing, a novel neurosymbolic approach that integrates deep learning with argumentation construction and reasoning for interpretable classification with different data modalities. In our approach deep neural networks construct an argumentation structure wherein data points support their assigned label and attack different ones. Using differentiable argumentation semantics for reasoning, the model is trained end-to-end to jointly learn feature representation and argumentative interactions. This results in argumentation structures providing faithful case-based explanations for predictions. Structure constraints over the argumentation graph guide learning, improving both interpretability and predictive performance. Experiments with tabular and imaging datasets show that Deep Arguing achieves performance competitive with standard baselines whilst offering interpretable argumentative reasoning.

Large reasoning models, such as OpenAI o1 and DeepSeek-R1, tend to become increasingly verbose as their reasoning capabilities improve. These inflated Chain-of-Thought (CoT) trajectories often exceed what the underlying problems require, wasting compute, latency, and context budgets. While introducing length-based efficiency rewards during reinforcement learning offers a natural remedy, existing methods struggle with two fundamental challenges: the optimal balance between correctness and efficiency is non-stationary throughout training, and intrinsic reasoning budgets vary drastically across problems. Relying on static reward weights and global length constraints inevitably forces a compromise between degraded accuracy and unrealized compression. To overcome these limitations, we propose LEAD (Length-Efficient Adaptive and Dynamic reasoning), a method that replaces static heuristics with online, self-adaptive mechanisms. LEAD dynamically calibrates the correctness-efficiency trade-off at each step using a Potential-Scaled Instability, directing optimization capacity to the most informative learning signal. Furthermore, it estimates an adaptive per-problem target length online based on the model's own correct rollouts, applying a symmetric efficiency reward that penalizes both overthinking and over-compression. Evaluated on five mathematical reasoning benchmarks, LEAD achieves the highest accuracy and Accuracy-Efficiency Score among RL-trained efficient-reasoning methods while producing substantially shorter outputs than the base model.

While explicit Chain-of-Thought (CoT) empowers large reasoning models (LRMs), it enables the generation of riskier final answers. Current alignment paradigms primarily rely on externally enforced compliance, optimizing models to detect malicious prompts rather than evaluating the safety of their own outputs. We argue that this approach remains largely behavioral: our empirical analysis reveals that ostensibly aligned models lack intrinsic safety understanding, often failing to verify their own response safety and remaining vulnerable to adversarial jailbreaks. To address this fundamental limitation, we propose Safety Internal (SInternal), a framework that internalizes safety specifications by training LRMs exclusively on safety verification tasks to critique their own generated answers using expert reasoning trajectories. We demonstrate that learning to verify induces a strong generalization for response safety, significantly enhancing robustness against out-of-domain jailbreaks. Furthermore, when combined with reinforcement learning, SInternal serves as a superior initialization compared to standard supervised fine-tuning, suggesting that internalizing safety understanding creates a more robust foundation for alignment than merely mimicking safe behaviors. Our codes are available at https://github.com/AlphaLab-USTC/SInternal

Large language model agents increasingly operate through environment-facing scaffolds that expose files, web pages, APIs, and logs. These observations influence tool use, state tracking, and action sequencing, yet their reliability and authority are often uncertain. Environmental grounding is therefore a systems-level problem involving context admission, evidence provenance, freshness checking, verification policy, action gating, and model reasoning. Existing agent benchmarks mainly evaluate task capability or specific attacks such as prompt injection and memory poisoning, but they under-specify a fundamental reliability question: whether agents remain grounded in the true environment state when observations are stale, incorrect, or malicious. We introduce EnvTrustBench, an agentic framework for benchmarking this failure mode. We define an evidence-grounding defect (EGD) as a behavioral failure in which an agent treats an environment-facing claim as sufficient evidence for action without resolving it against available current evidence, leading to a task-incorrect false path under the true environment state. Given a task scenario, EnvTrustBench generates the workspace, environment, agent-facing objective, and validation oracle, executes the evaluated agent, records its action-observation trajectory and final state, and applies the oracle to produce a verdict. Using 6 LLM backbones and 5 widely used scaffolds, we evaluate 55 generated cases across 11 task scenarios, with each scenario expanded through five feedback-guided generation iterations. Results show that EGDs consistently emerge across operational workflows, highlighting environmental grounding as a core agent reliability problem with important security implications.

Large reasoning models achieve high accuracy through extended chain-of-thought but generate 5--8 more tokens than necessary, applying verbose reasoning uniformly regardless of problem difficulty. We propose Hint Tuning, a data-efficient approach that teaches models to calibrate reasoning depth. Our key insight: the corresponding instruct model serves as an ideal difficulty probe. By testing what the instruct model can solve with varying guidance, we automatically construct training data across three states: No-Hint (direct answer), Sparse-Hint (minimal prefix), and Full-Hint (complete reasoning). This converts the abstract challenge of difficulty labeling into a measurable consistency check between the instruct and reasoning models. With only 1K self-annotated samples, Hint Tuning achieves 24--66% token reduction (31.5% average) across mainstream reasoning models (Qwen3-Thinking, DeepSeek-R1-Distill) at multiple scales (4B--32B) while maintaining competitive accuracy on five benchmarks. Unlike methods requiring massive distillation datasets or expensive RL, we achieve superior efficiency through simple alignment with the instruct model's capabilities. Code and data are available at https://github.com/redai-infra/hint-tuning.

Humans rapidly learn abstract knowledge when encountering novel environments and flexibly deploy this knowledge to guide efficient and intelligent action. Can modern AI systems learn and plan in a similar way? We study this question using a dataset of complex human gameplay with concurrent fMRI recordings, in which participants learn novel video games that require rule discovery, hypothesis revision, and multi-step planning. We jointly evaluate models by their ability to play the games, match human learning behavior, and predict brain activity during the same task, comparing a suite of frontier Large Reasoning Models (LRMs) against model-free and model-based deep reinforcement learning agents and a Bayesian theory-based agent. We find that frontier LRMs most closely match human behavioral patterns during game discovery and predict brain activity an order of magnitude better than both reinforcement learning alternatives across cortical and subcortical regions, with effects robust to permutation controls. Through targeted manipulations, we further show that brain alignment reflects the model's in-context representation of the game state rather than its downstream planning or reasoning. Our results establish LRMs as compelling computational accounts of human learning and decision making in complex, naturalistic environments. Project page with interactive replays: https://botcs.github.io/reason-to-play/

Despite the rapid development of AI reviewers, evaluating such systems remains challenging: metrics favor overlap with human reviews over correctness. However, since human reviews often cover only a subset of salient issues and sometimes contain mistakes, they are unreliable as gold references. To address this, we build category-specific benchmark subsets and skip evaluation when the corresponding human reviews are missing to strengthen Completeness. We also leverage reviewer--author--meta-review discussions as expert annotations and filter unreliable reviews accordingly to strengthen Correctness. Finally, we introduce CoCoReviewBench, which curates 3,900 papers from ICLR and NeurIPS to enable reliable and fine-grained evaluation of AI reviewers. Analysis shows that AI reviewers remain limited in correctness and are prone to hallucinations, and highlights reasoning models as more effective reviewers, motivating further directions for improving AI reviewers. Benchmarks and models are available at https://github.com/hexuandeng/CoCoReviewBench.

Large reasoning models (LRMs) achieve strong performance via extended chain-of-thought (CoT) reasoning, yet suffer from excessive token consumption and high inference latency. Existing reinforcement learning (RL) approaches for CoT compression rely on uniform, static length penalties that neglect model capability dynamics and problem-level difficulty variation. We propose \textbf{ExpThink}\xspace, an RL framework that addresses both dimensions through two complementary mechanisms. First, \emph{experience-guided reward shaping} tracks the shortest correct solution found so far for each problem and applies a three-tier reward: full credit for concise correct responses, discounted credit for verbose correct ones, and zero for incorrect ones. The threshold tightens automatically with model improvement, forming a self-evolving curriculum that requires no manual scheduling. Second, \emph{difficulty-adaptive advantage} replaces standard deviation normalization with correct-count normalization, yielding monotonically difficulty-scaled gradients that amplify learning on hard problems to preserve accuracy while suppressing gradients on easy ones to encourage brevity. Together, these mechanisms enforce an accuracy-first, compression-second training objective. Experiments on multiple mathematical reasoning benchmarks demonstrate that \textbf{ExpThink}\xspace reduces average response length by up to 77% while simultaneously improving accuracy, achieving up to $3\times$ higher accuracy-efficiency ratio (accuracy divided by average token count) than the vanilla baseline and outperforming existing RL-based compression methods on both metrics.

Large Language Models (LLMs) have achieved great improvements in recent years. Nevertheless, it still remains unclear how good LLMs are for reasoning tasks, especially for long-chain ones. In this paper, we evaluate LLMs' performance on the simplest yet long-chain reasoning task, namely the Equivalence Class Problem (ECP), i.e., determining whether two variables are equal given a set of randomly generated equivalence relations. We consider both reasoning and non-reasoning representative LLMs over a large variety of problem instances, ranging over different numbers of variables, connectivity probabilities, prompts, and other factors. The experimental results show that non-reasoning LLMs fail ECP, while reasoning models are significantly better but still struggle to completely solve this problem. Interestingly, considering various connectivity probabilities with a fixed number of variables, we observe that, for non-reasoning models, the hardest problem instances coincide with the phase transition point of ln n/(n-1), suggesting the chaos of the problem; in contrast, for reasoning models, the hardest ones coincide with the biggest diameter, suggesting the reasoning difficulty of the problem.

Large language models (LLMs), especially reasoning models, generate extended chain-of-thought (CoT) reasoning that often contains explicit deliberation over future outcomes. Yet whether this deliberation constitutes genuine planning, how it is structured, and what aspects of it drive performance remain poorly understood. In this work, we introduce a new method to characterize LLM planning by extracting and quantifying search trees from reasoning traces in the four-in-a-row board game. By fitting computational models on the extracted search trees, we characterize how plans are structured and how they influence move decisions. We find that LLMs' search is shallower than humans', and that performance is predicted by search breadth rather than depth. Most strikingly, although LLMs expand deep nodes in their traces, their move choices are best explained by a myopic model that ignores those nodes entirely. A causal intervention study where we selectively prune CoT paragraphs further suggests that move selection is driven predominantly by shallow rather than deep nodes. These patterns contrast with human planning, where performance is driven primarily by deep search. Together, our findings reveal a key difference between LLM and human planning: while human expertise is driven by deeper search, LLMs do not act on deep lookahead. This dissociation offers targeted guidance for aligning LLM and human planning. More broadly, our framework provides a generalizable approach for interpreting the structure of LLM planning across strategic domains.

On-Policy Self-Distillation (OPSD) has recently emerged as an alternative to Reinforcement Learning with Verifiable Rewards (RLVR), promising higher accuracy and shorter responses through token-level credit assignment from a self-teacher conditioned on privileged context. However, this promise does not carry over to thinking-enabled mathematical reasoning, where reported accuracy gains shrink and sometimes turn negative. We hypothesize that hindsight supervision can specify better token-level alternatives in short thinking-disabled outputs, but in long thinking-enabled traces it more readily identifies redundancy than supplies better replacements. To test this, we applied OPSD separately to correct and incorrect rollout groups, so that compression and correction can be observed in isolation. Our results show that in thinking-enabled mathematical reasoning, OPSD behaves most reliably as a compression mechanism rather than a correction mechanism: training only on correct rollouts preserves accuracy while substantially shortening responses, whereas training only on incorrect rollouts damages accuracy. In light of these findings, we propose a revised post-training pipeline for thinking-enabled mathematical reasoning: SFT then RLVR then OPSD.

Large Language Models (LLMs) are evolving into autonomous trading agents, yet existing benchmarks often overlook the interplay between architectural reasoning and strategy consistency. We propose Strat-LLM, a framework grounded in Stratified Strategy Alignment. Operating in a live-forward setting throughout 2025, it integrates heterogeneous data including sequential prices, real-time news, and annual reports to eliminate look-ahead bias. Extensive stress tests on A-share and U.S. markets reveal: (1) reasoning-heavy models achieve peak utility in Free Mode via internal logic, whereas standard models require Strict Mode as a vital risk anchor; (2) alignment utility is regime-dependent, with Free and Guided modes capturing momentum in uptrending markets, while Strict Mode mitigates drawdowns in downtrends; (3) mid-scale models (35B) show optimal fidelity under strict constraints, whereas ultra-large models (122B) suffer an alignment tax under rigid rules but gain a performance premium in Guided Mode; (4) standard LLMs often fall into a high win-rate trap, optimizing for small gains at the expense of total returns, which can only be mitigated through deep reasoning or strict external guardrails. Project details are available at https://Strat-LLM.github.io.

Large reasoning models (LRMs) increasingly expose chain-of-thought-like reasoning for transparency, verification, and deliberate problem solving. This creates a safety blind spot: harmful or policy-violating content may appear in reasoning traces even when final answers appear safe. We test whether final-answer safety is a sufficient proxy for the full reasoning-answer trajectory by scoring both stages under a unified twenty-principle safety rubric. Using prompts from seven public harmfulness and jailbreak sources, plus four out-of-distribution (OOD) sources, we evaluate 15 open-weight and API-based LRMs across 41K prompts per model. Reasoning traces consistently reveal additional safety risks beyond final answers, especially in high-severity stage-wise failures: leak cases, where unsafe reasoning precedes a safe-looking answer, and escape cases, where benign-looking reasoning precedes an unsafe final response. Principle-level analysis shows that risk concentrates in misinformation, legal compliance, discrimination, physical harm, and psychological harm. We further propose adaptive multi-principle steering, a white-box test-time mitigation that learns one unsafe-to-safe activation direction per safety principle and activates only directions whose current hidden state is closer to the unsafe than safe centroid. On three steerable open reasoning models, adaptive steering reduces unsafe counts in both reasoning traces and final answers on held-out and OOD benchmarks. DeepSeek-R1-Qwen-7B achieves a 40.8% average unsafe-count reduction while retaining 97.7% macro-averaged accuracy on BBH, GSM8K, and MMLU. These results suggest that LRM safety should be evaluated and mitigated over the full exposed reasoning-answer trajectory, not only at the final-answer stage.

Computational cognitive models discovered using large language models have so far relied solely on behavioral data. However, it is well-known that models produced from the behavioral trajectory alone are typically under-determined. In this work, we explore the use of Think Aloud traces as an additional form of data constraint during automated model discovery. When applied to the domain of risky decision-making, we find that the models discovered with think-aloud achieve significantly improved predictive performance on held-out data. Additionally, we find that the discovered models belong to different structural classes than those discovered from behavior alone for the majority of participants (69.4%), specifically, it shifts from Explicit comparator towards Integrated utility. These results suggest that process-level language data not only improve model fit, but also systematically reshape the structure of the discovered cognitive models, enabling the identification of mechanisms that are not recoverable from behavior alone.

This paper presents the Personalized Thinking Model (PTM), a hierarchical and interpretable learner representation designed for AI supported education. PTM organizes evidence from learner journals into a five-layer structure covering behavioral instances, behavioral patterns, cognitive routines, metacognitive tendencies, and self-system values. PTM is grounded in Marzano's New Taxonomy of Educational Objectives and tries to clone learner's thinking model and build cognitive twin. It was constructed using a pipeline that combines large language model inference (Gemini 2.5 Pro), sentence embeddings, dimensionality reduction, and consensus clustering. This paper evaluates PTM fidelity through three methods applied to 40 participants in a seven-week study. First, automatic evaluation using atomic information point matching yielded an overall F1 score of 74.57% before human-in-the-loop (HITL) refinement and 75.48% after refinement. Second, user evaluation using a Likert scale produced mean ratings of 4.26 and 4.30 on a five-point scale for pre and post-HITL conditions respectively. Third, semantic alignment verification showed that topic coherence increased from 0.436 at the behavioral layer to 0.626 at the core value layer, while lexical overlap with journal vocabulary decreased from 0.114 to 0.007 across those same layers. These results suggest that the PTM produces outputs with acceptable fidelity, was generally perceived by users as reflecting their thinking, and showed a pattern consistent with semantic abstraction across layers.

Cyber threats are rapidly increasing, expanding their impact from large-scale enterprises to government services and individual users, making robust security systems increasingly essential. However, a significant shortage of skilled cybersecurity professionals exacerbates this challenge. While recent research has explored automating tasks such as penetration testing using LLM-based agents, existing frameworks often perform poorly due to limited capability in strategy formulation, domain-specific reasoning, and accurate action and tool selection. To overcome these limitations, we propose Pen-Strategist framework, consisting of a novel domain-specific reasoning model that derives pentesting strategies via logical reasoning and a classifier that converts the strategies into actionable steps. First, we construct a reasoning dataset containing logical explanations for both strategy derivation and step selection in pentesting scenarios. We then fine-tune a Qwen-3-14B model for strategy generation using reinforcement learning. Evaluation on the test split of the dataset demonstrates a 87% improvement in strategy derivation performance compared to the baseline. Furthermore, we integrate the fine-tuned Pen-Strategist model into existing automated pentesting frameworks, such as PentestGPT, and evaluate its performance on vulnerable machines, achieving a 47.5% improvement in subtask completion while surpassing the baseline GPT-5. Further experiments on the CTFKnow benchmark show an 18% performance gain over the base model. For step prediction, we train a semantic-based CNN classifier, which outperforms commercial LLMs by 28% and enhances execution stability. Finally, we conduct a user study to qualitatively assess the generated strategies, and Pen-Strategist demonstrates superior performance compared to the Claude-4.6-Sonnet.

Recent advances in Large Reasoning Models (LRMs) trained with Long Chain-of-Thought have demonstrated remarkable capabilities in code generation and mathematical reasoning. However, their potential in multi-turn Text-to-SQL tasks remains largely underexplored. Existing approaches typically rely on unstable API-based inference or require expensive fine-tuning on small-scale models. In this work, we present Rose-SQL, a training-free framework that leverages small-scale LRMs through in-context learning to enable accurate context-dependent parsing. We introduce the Role-State, a fine-grained representation that bridges the structural gap between schema linking and SQL generation by serving as a structural blueprint. To handle conversational dependencies, Rose-SQL traces the evolution of Role-State through historical context via structural isomorphism checks, guiding the model to infer the possible SQL composition for the current question through verified interaction trajectories. Experiments on the SParC and CoSQL benchmarks show that, within the Qwen3 series, Rose-SQL outperforms in-context learning baselines at the 4B scale and substantially surpasses state-of-the-art fine-tuned models at the 8B and 14B scales, while showing consistent gains on additional reasoning backbones.

Time series (TS) reasoning models (TSRMs) have shown promising capabilities in general domains, yet they consistently fail in the financial domain, which exhibits unique characteristics. We propose a general 2 x 2 capability taxonomy for TSRMs by crossing 1) single-entity vs. multi-entity analysis with 2) assessment of the current state vs. prediction of future behavior. We instantiate this taxonomy in the financial domain-where the distinction between deterministic assessment and stochastic prediction is particularly critical-as ten financial reasoning tasks, forming the FinTSR-Bench benchmark based on S&P stocks. To this end, we propose FinSTaR (Financial Time Series Thinking and Reasoning), trained on FinTSR-Bench with distinct chain-of-thought (CoT) strategies tailored to each category. For assessment, which is deterministic (i.e., computable from observable data), we employ Compute-in-CoT, a programmatic CoT that enables models to derive answers directly from raw prices. For prediction, which is inherently stochastic (i.e., subject to unobservable factors), we adopt Scenario-Aware CoT, which generates diverse scenarios before making a judgment, mirroring how financial analysts reason under uncertainty. The proposed method achieves 78.9% average accuracy on FinTSR-Bench, substantially outperforming LLM and TSRM baselines. Furthermore, we show that the four capability categories are complementary and mutually reinforcing through joint training, and that Scenario-Aware CoT consistently improves prediction accuracy over standard CoT. Code is available at https://github.com/seunghan96/FinSTaR.

Millions of users turn to AI models for their information needs. It is conceivable that a large number of user queries contain assumptions that may be factually inaccurate. Prior work notes that large language models (LLMs) often fail to challenge such erroneous assumptions, and can reinforce users' misinformed opinions. However, given the recent advances, especially in model's reasoning capabilities, we revisit whether large reasoning models (LRMs) can reason about the underlying assumptions and respond to user queries appropriately. We construct queries with varying degrees of presuppositions spanning health, science, and general knowledge, and use it to evaluate several widely-deployed models When compared to non-reasoning models, we find that reasoning models achieve a slightly higher accuracy (2-11%), but they still fail to challenge a large fraction (26-42%) of false presuppositions. Further, reasoning models remain susceptible to how strongly the presupposition is expressed.

Vision-Language-Action (VLA) models aim to provide a single generalist controller for robots, but today's systems fall short on the criteria that matter for real-world deployment. Frontier models are closed, open-weight alternatives are tied to expensive hardware, reasoning-augmented policies pay prohibitive latency for their grounding, and fine-tuned success rates remain below the threshold for dependable use. We present MolmoAct2, a fully open action reasoning model built for practical deployment, advancing its predecessor along five axes. We introduce MolmoER, a VLM backbone specialized for spatial and embodied reasoning, trained on a 3.3M-sample corpus with a specialize-then-rehearse recipe. We release three new datasets spanning low-to-medium cost platforms, including MolmoAct2-BimanualYAM, 720 hours of teleoperated bimanual trajectories that constitute the largest open bimanual dataset to date, together with quality-filtered Franka (DROID) and SO100/101 subsets. We provide OpenFAST, an open-weight, open-data action tokenizer trained on millions of trajectories across five embodiments. We redesign the architecture to graft a flow-matching continuous-action expert onto a discrete-token VLM via per-layer KV-cache conditioning. Finally, we propose MolmoThink, an adaptive-depth reasoning variant that re-predicts depth tokens only for scene regions that change between timesteps, retaining geometric grounding at a fraction of prior latency. In the most extensive empirical study of any open VLA to date, spanning 7 simulation and real-world benchmarks, MolmoAct2 outperforms strong baselines including Pi-05, while MolmoER surpasses GPT-5 and Gemini Robotics ER-1.5 across 13 embodied-reasoning benchmarks. We release model weights, training code, and complete training data. Project page: https://allenai.org/blog/molmoact2

Legal texts often contain computational legal clauses--provisions whose understanding requires complex logic. While frontier Large Reasoning Models (LRMs) can describe such clauses, building production-ready systems is limited by reasoning errors and the high cost of inference. We propose Amortized Intelligence, a neuro-symbolic approach where we use an LLM once to translate a legal text into Deterministic Autonomous Contract Language (DACL): a typed graph intermediate representation. Adjudication then relies on deterministic graph executions with a visually auditable trace. In comparison against runtime LRM baselines (including GPT-5.2 and Gemini 3 Pro), our DACL-based Agent achieves near-perfect consistency and mitigates the "reasoning cliff" observed in probabilistic models. The system reduces compute costs by over 90% in high-volume workflows while satisfying the strict auditability requirements of legal adjudication.

Specification gaming is a critical failure mode of LLM agents. Despite this, there has been little systematic research into when it arises and what drives it. To address this, we build and open source a diverse suite of tasks where models can score highly by taking unintended actions. We find that all tested models exploit their specifications at non-negligible rates in most of our eight settings, including five non-coding settings. We see the highest rates of specification gaming in Grok 4 and the lowest rates in Claude models. We use our evaluation suite to study what drives specification gaming, and find that: 1. RL reasoning training substantially increases the rate at which models exploit their specifications, 2. Increasing RL reasoning budget has a weakly positive effect on exploit rate, and 3. Test-time mitigations reduce but do not eliminate the rate of specification gaming. Our results suggest that specification gaming is a fundamental challenge arising from RL reasoning training; we release our evaluation suite to support further work on this problem.

Multimodal large language models (MLLMs) have shown considerable potential in chart understanding and reasoning tasks. However, they still struggle with high information density (HID) charts characterized by multiple subplots, legends, and dense annotations due to three major challenges: (1) limited fine-grained perception results in the omission of critical visual cues; (2) redundant or noisy visual information undermines the performance of multimodal reasoning; (3) lack of adaptive deep reasoning relative to the amount of visual information. To tackle these challenges, we present a novel focus-driven fine-grained chart reasoning model, Chart-FR1, to improve perception, focusing efficiency, and adaptive deep reasoning on HID charts. Specifically, we propose Focus-CoT, a visual focusing chain-of-thought that enhances fine-grained perception by explicitly linking reasoning steps to key visual cues, such as local image regions and OCR signals. Building on this, we introduce Focus-GRPO, a focus-driven reinforcement learning algorithm with an information-efficiency reward that compresses redundant visual information for efficient focusing, and an adaptive KL penalty mechanism that enables flexible control over reasoning depth as more visual cues are discovered. Furthermore, to fill the gap in benchmarks for HID charts, we build HID-Chart, a challenging benchmark with an information-density metric designed to evaluate fine-grained chart reasoning capabilities. Extensive experiments on multiple chart benchmarks demonstrate that Chart-FR1 outperforms state-of-the-art MLLMs in chart understanding and reasoning. Code is available at https://github.com/phkhub/Chart-FR1.

Large Language Models (LLMs) have substantially advanced the field of Natural Language Processing (NLP), achieving state-of-the-art performance across a wide range of tasks. These improvements have been attributed, in part, to their emerging reasoning capabilities, which are enabled by large-scale training and increased model capacity. However, existing LLMs can generate erroneous responses when addressing complex queries that fall outside their training distribution, due to limited internal knowledge or the need for multi-step reasoning. To address these limitations, recent work has introduced large reasoning models (LRMs), which incorporate explicit internal reasoning processes to improve response accuracy. Additionally, state-of-the-art LRMs often comprise hundreds of billions of parameters and require several seconds per inference, even on advanced multi-GPU systems. These characteristics limit their practicality for deployment in conventional computing environments. Meanwhile, NLP research on multilingual LLMs continues to prioritize high-resource languages. However, these models exhibit limited performance in under-resourced languages, primarily due to insufficient language- and culture-specific training data. In this paper, we focus on Modern Greek, for which only a limited number of question answering (QA) datasets have been proposed, most of which are intended for model evaluation. To address this research gap in Greek QA, we make the following contributions: (i) CulturaQA, a high-quality LRM-generated and human-curated dataset, for Greek LLM training and evaluation; (ii) a memory-efficient LLM evaluation framework adaptable to diverse languages and QA tasks; (iii) Maistros 8B, a state-of-the-art open-weights Greek LLM developed via knowledge distillation and fine-tuning on CulturaQA; and (iv) a comprehensive evaluation of nine LLMs across nine human-curated Greek QA datasets.

Large Reasoning Models achieve strong performance on complex tasks but remain prone to hallucinations, particularly in long-form generation where errors compound across reasoning steps. Existing approaches to improving factuality, including abstention and factuality-driven optimization, follow a \emph{coupled exploration-commitment} paradigm, in which intermediate reasoning is unconditionally propagated to the final output, limiting fine-grained control over information selection and integration. In this paper, we propose an \textbf{Exploration-Commitment Decoupling} paradigm that disentangles knowledge exploration from final commitment, enabling models to explore with awareness while answering cautiously. We instantiate the paradigm with \textbf{Calibration-Aware Generation (CAG)}, a framework that equips models with end-to-end, calibration-aware generation capabilities, by augmenting intermediate reasoning with calibrated reliability estimates and prioritizing reliable content in final outputs. Across five long-form factuality benchmarks and multiple model families, CAG improves factuality by up to 13%, while reducing decoding time by up to 37%. Overall, our work highlights decoupling as a principled approach for more reliable long-form generation, offering directions for trustworthy and self-aware generative systems.

The inception of Large Language Models (LLMs) has catalyzed AI adoption in the finance sector, yet their reliability in complex, jurisdiction-specific tasks like Indian Chartered Accountancy (CA) remains limited. The models display difficulty in executing numerical tasks which require multiple steps while also needing advanced knowledge about legal regulations and the method of scaling their operations is not feasible in settings which have limited access to resources. We present CA-ThinkFlow as a parameter-efficient Retrieval-Augmented Generation (RAG) framework which operates with a 14B, 4-bit-quantized reasoning model, 14B-DeepSeek-R1, and a layout-aware Docling extraction system which maintains document structure during extraction. CA-ThinkFlow uses a basic RAG method which automatically adds retrieved information into the prompt, while it depends on the model's built-in Chain-of-Thought (CoT) functions to create context and produce correct answers. The system we developed system operates at performance levels which match large proprietary models when we tested it on the multi-level CA-Ben benchmark, achieving Scholastic Reliability Coefficient (SRC) results which equal 68.75% of GPT-4o and Claude 3.5 Sonnet. The framework shows high efficiency and strength in handling parameters, but essential reasoning abilities fail to process complex regulatory texts which exist in fields such as Taxation.

Explainable AI (XAI) aims to improve user understanding and decisions when using AI models. However, despite innovations in XAI, recent user evaluations reveal that this goal remains elusive. Understanding human cognition can help explain why users struggle to effectively use AI explanations. Focusing on reasoning on structured (tabular) data, we examined various reasoning strategies for different XAI methods (none, feature importance, feature attribution) in the decision task of anticipating AI decisions (i.e., forward simulation). We i) elicited reasoning strategies from a formative user study, and ii) collected decisions from a summative user study. Using cognitive modeling, we implemented the processes underlying each reasoning strategy and evaluated their alignment with human decision-making. We found that our models better fit human decisions than baseline machine learning proxies, providing insights into which reasoning strategies are (in)effective. We then demonstrate how the fitted model can be used to form hypotheses and investigate research questions that are costly to study with real human participants. This work contributes to debugging human understanding of XAI, informing the future development of more usable and interpretable AI explanations.

Tool-calling agents are evaluated on tool selection, parameter accuracy, and scope recognition, yet LLM trajectory assessments remain inherently post-hoc. Disconnected from the active execution loop, such assessments identify errors that are usually addressed through prompt-tuning or retraining, and fundamentally cannot course-correct the agent in real time. To close this gap, we move evaluation into the execution loop at inference time: a specialized reviewer agent evaluates provisional tool calls prior to execution, shifting the paradigm from post-hoc recovery to proactive evaluation and error mitigation. In practice, this architecture establishes a clear separation of concerns between the primary execution agent and a secondary review agent. As with any multi-agent system, the reviewer can introduce new errors while correcting others, yet no prior work to our knowledge has systematically measured this tradeoff. To quantify this tradeoff, we introduce Helpfulness-Harmfulness metrics: helpfulness measures the percentage of base agent errors that feedback corrects; harmfulness measures the percentage of correct responses that feedback degrades. These metrics directly inform reviewer design by revealing whether a given model or prompt provides net positive value. We evaluate our approach on BFCL (single-turn) and Tau2-Bench (multi-turn stateful scenarios), achieving +5.5% on irrelevance detection and +7.1% on multi-turn tasks. Our metrics reveal that reviewer model choice is critical: the reasoning model o3-mini achieves a 3:1 benefit-to-risk ratio versus 2.1:1 for GPT-4o. Automated prompt optimization via GEPA provides an additional +1.5-2.8%. Together, these results demonstrate a core advantage of separating execution and review: the reviewer can be systematically improved through model selection and prompt optimization, without retraining the base agent.

Large reasoning models such as DeepSeek-R1 and OpenAI o1 generate extended chains of thought spanning thousands of tokens, yet their integration with retrieval-augmented generation (RAG) remains fundamentally misaligned. Current RAG systems optimize for providing context before reasoning begins, while reasoning models require evidence injection during multi-step inference chains. We introduce ReaLM-Retrieve, a reasoning-aware retrieval framework that addresses this mismatch through three key innovations: (1) a step-level uncertainty detector that identifies knowledge gaps at reasoning-step granularity rather than token or sentence level; (2) a retrieval intervention policy that learns when external evidence maximally benefits ongoing reasoning; and (3) an efficiency-optimized integration mechanism that reduces per-retrieval overhead by 3.2x compared to naive integration. Experiments on MuSiQue, HotpotQA, and 2WikiMultiHopQA demonstrate that ReaLM-Retrieve achieves on average 10.1% absolute improvement in answer F1 over standard RAG (range: 9.0-11.8% across the three benchmarks) while reducing retrieval calls by 47% compared to fixed-interval approaches like IRCoT (all improvements significant at p<0.01, paired bootstrap). On the challenging MuSiQue benchmark requiring 2-4 hop reasoning, our method achieves 71.2% F1 with an average of only 1.8 retrieval calls per question. Analysis shows that ReaLM-Retrieve also improves retrieval quality itself, achieving 81.3% Recall@5 with consistently higher precision and MRR than fixed-interval baselines on supporting evidence, establishing new state-of-the-art efficiency-accuracy trade-offs for reasoning-intensive retrieval tasks.

Large Reasoning Models (LRMs) achieve strong performance on mathematical reasoning tasks but remain unreliable on challenging instances. Existing test-time scaling methods, such as repeated sampling, self-correction, and tree search, improve performance at the cost of increased computation, yet often exhibit diminishing returns on hard problems. We observe that output disagreement is strongly correlated with instance difficulty and prediction correctness, providing a useful signal for guiding instance-level strategy selection at test time. Based on this insight, we propose a training-free framework that formulates test-time scaling as an instance-level routing problem, rather than allocating more computation within a single strategy, dynamically selecting among different scaling strategies based on output disagreement. The framework applies lightweight resolution for consistent cases, majority voting for moderate disagreement, and rewriting-based reformulation for highly ambiguous instances. Experiments on seven mathematical benchmarks and three models show that our method improves accuracy by 3% - 7% while reducing sampling cost compared to existing approaches.

SFT-then-RLVR is widely used for post-training reasoning models, but why this specific ordering, and why RLVR-only stalls at cold start, have lacked a unifying theoretical account. We provide that account under a unified loss family $J_Q$ using the Tsallis $q$-logarithm. $J_Q$ is a single-parameter family that interpolates between RLVR (at $q{=}0$, the \textit{exploitation pole}) and the log-marginal-likelihood over latent trajectories (at $q{=}1$, the \textit{density-estimation pole}), under which the standard pipeline corresponds to a stepwise $q{=}1 \to 0$ schedule. All members share the same per-example gradient direction, differing only by a per-instance amplification $P_θ^{-q}$ that reweights each instance independently of the learning rate. Under gradient flow analysis, we show that the exploitation pole requires $Ω(\frac{1}{p_0})$ time to escape cold start but is robust to label noise, while the density-estimation pole escapes in $Θ\big(\log(\frac{1}{p_0})\big)$ but memorizes label noise. This separation explains how SFT ($q{=}1$) first moves the model out of the cold-start regime, followed by the more robust RLVR ($q{=}0$), under the SFT-then-RLVR paradigm. We further derive two Monte Carlo estimators that directly optimize fixed-$q$ on the $J_Q$ continuum, without annotated rationales: Gradient-Amplified RL (GARL) and Posterior-Attenuated Fine-Tuning (PAFT), with shared bias $O\big(\frac{q}{M P_θ^q}\big)$ but different variance and stability properties. On FinQA, HotPotQA, and MuSiQue, GARL at sufficiently high $q$ substantially mitigates cold-start stalling, escaping cold start where GRPO fails entirely. In warm start, GARL at low $q$ dominates FinQA where training is stable; on HotPotQA and MuSiQue, GARL destabilizes and PAFT at $q{=}0.75$ remains stable, reaching $47.9$ \texttt{m@16} on HotPotQA ($+13.9$ over GRPO).