WorldMemArena: Evaluating Multimodal Agent Memory Through Action-World Interaction

Multimodal large language models are increasingly deployed as long-horizon agents, where memory must do more than recall: it must track an evolving world, revise what has gone stale, and surface the right evidence at decision time. Existing benchmarks measure recall over static dialogue, collapse memory into a single end-of-task accuracy, and reduce visual observations to captions, leaving us unable to localize failures to writing, maintenance, retrieval, or use. The rise of agent harnesses that author their own memory sharpens this gap, since we have no principled way to compare hand-designed pipelines with self-managing alternatives. To close these gaps, we formulate multimodal agent memory as an Action-World Interaction Loop with an observable four-stage lifecycle, and instantiate it in WorldMemArena: 400 multi-session multimodal tasks spanning Lifelong Evolution (evolving personal and task states) and Agentic Execution (memory from real observations, actions, and feedback), annotated with gold memory points, updates, distractors, and evidence chains for stage-level diagnosis. This enables the first head-to-head comparison of long-context, manually designed (RAG and external memory systems), and harness-based memory agents. Results show that: (1) better memory writing and storage do not guarantee better performance; (2) multimodal memory still struggles to fully use visual evidence; (3) systems are unstable across domains and degrade on realistic agentic trajectories; and (4) harness memory is more flexible but remains costly and less reliable.

Grounded or Guessing? LVLM Confidence Estimation via Blind-Image Contrastive Ranking

Large vision-language models suffer from visual ungroundedness: they can produce a fluent, confident, and even correct response driven entirely by language priors, with the image contributing nothing to the prediction. Existing confidence estimation methods cannot detect this, as they observe model behavior under normal inference with no mechanism to determine whether a prediction was shaped by the image or by text alone. We introduce BICR (Blind-Image Contrastive Ranking), a model-agnostic confidence estimation framework that makes this contrast explicit during training by extracting hidden states from a frozen LVLM twice: once with the real image-question pair, and once with the image blacked out while the question is held fixed. A lightweight probe is trained on the real-image hidden state and regularized by a ranking loss that penalizes higher confidence on the blacked-out view, teaching it to treat visual grounding as a signal of reliability at zero additional inference cost. Evaluated across five modern LVLMs and seven baselines on a benchmark covering visual question answering, object hallucination detection, medical imaging, and financial document understanding, BICR achieves the best cross-LVLM average on both calibration and discrimination simultaneously, with statistically significant discrimination gains robust to cluster-aware analysis at 4-18x fewer parameters than the strongest probing baseline.

Deep FinResearch Bench: Evaluating AI's Ability to Conduct Professional Financial Investment Research

We introduce Deep FinResearch Bench, a practical and comprehensive evaluation framework for deep research (DR) agents in financial investment research. The benchmark assesses three dimensions of report quality: qualitative rigor, quantitative forecasting and valuation accuracy, and claim credibility and verifiability. Particularly, we define corresponding qualitative and quantitative evaluation metrics and implement an automated scoring procedure to enable scalable assessment. Applying the benchmark to financial reports from frontier DR agents and comparing them with reports authored by financial professionals, we find that AI-generated reports still fall short across these dimensions. These findings underscore the need for domain-specialized DR agents tailored to finance, and we hope the work establishes a foundation for standardized benchmarking of DR agents in financial research.

Distill and Align Decomposition for Enhanced Claim Verification

Complex claim verification requires decomposing sentences into verifiable subclaims, yet existing methods struggle to align decomposition quality with verification performance. We propose a reinforcement learning (RL) approach that jointly optimizes decomposition quality and verifier alignment using Group Relative Policy Optimization (GRPO). Our method integrates: (i) structured sequential reasoning; (ii) supervised finetuning on teacher-distilled exemplars; and (iii) a multi-objective reward balancing format compliance, verifier alignment, and decomposition quality. Across six evaluation settings, our trained 8B decomposer improves downstream verification performance to (71.75%) macro-F1, outperforming prompt-based approaches ((+1.99), (+6.24)) and existing RL methods ((+5.84)). Human evaluation confirms the high quality of the generated subclaims. Our framework enables smaller language models to achieve state-of-the-art claim verification by jointly optimising for verification accuracy and decomposition quality.

How Reliable are Confidence Estimators for Large Reasoning Models? A Systematic Benchmark on High-Stakes Domains

The miscalibration of Large Reasoning Models (LRMs) undermines their reliability in high-stakes domains, necessitating methods to accurately estimate the confidence of their long-form, multi-step outputs. To address this gap, we introduce the Reasoning Model Confidence estimation Benchmark (RMCB), a public resource of 347,496 reasoning traces from six popular LRMs across different architectural families. The benchmark is constructed from a diverse suite of datasets spanning high-stakes domains, including clinical, financial, legal, and mathematical reasoning, alongside complex general reasoning benchmarks, with correctness annotations provided for all samples. Using RMCB, we conduct a large-scale empirical evaluation of over ten distinct representation-based methods, spanning sequential, graph-based, and text-based architectures. Our central finding is a persistent trade-off between discrimination (AUROC) and calibration (ECE): text-based encoders achieve the best AUROC (0.672), while structurally-aware models yield the best ECE (0.148), with no single method dominating both. Furthermore, we find that increased architectural complexity does not reliably outperform simpler sequential baselines, suggesting a performance ceiling for methods relying solely on chunk-level hidden states. This work provides the most comprehensive benchmark for this task to date, establishing rigorous baselines and demonstrating the limitations of current representation-based paradigms.

A Variational Approach for Mitigating Entity Bias in Relation Extraction

Mitigating entity bias is a critical challenge in Relation Extraction (RE), where models often rely excessively on entities, resulting in poor generalization. This paper presents a novel approach to address this issue by adapting a Variational Information Bottleneck (VIB) framework. Our method compresses entity-specific information while preserving task-relevant features. It achieves state-of-the-art performance on relation extraction datasets across general, financial, and biomedical domains, in both indomain (original test sets) and out-of-domain (modified test sets with type-constrained entity replacements) settings. Our approach offers a robust, interpretable, and theoretically grounded methodology.

GenPlanX. Generation of Plans and Execution

Classical AI Planning techniques generate sequences of actions for complex tasks. However, they lack the ability to understand planning tasks when provided using natural language. The advent of Large Language Models (LLMs) has introduced novel capabilities in human-computer interaction. In the context of planning tasks, LLMs have shown to be particularly good in interpreting human intents among other uses. This paper introduces GenPlanX that integrates LLMs for natural language-based description of planning tasks, with a classical AI planning engine, alongside an execution and monitoring framework. We demonstrate the efficacy of GenPlanX in assisting users with office-related tasks, highlighting its potential to streamline workflows and enhance productivity through seamless human-AI collaboration.

Conservative Bias in Large Language Models: Measuring Relation Predictions

Large language models (LLMs) exhibit pronounced conservative bias in relation extraction tasks, frequently defaulting to No_Relation label when an appropriate option is unavailable. While this behavior helps prevent incorrect relation assignments, our analysis reveals that it also leads to significant information loss when reasoning is not explicitly included in the output. We systematically evaluate this trade-off across multiple prompts, datasets, and relation types, introducing the concept of Hobson's choice to capture scenarios where models opt for safe but uninformative labels over hallucinated ones. Our findings suggest that conservative bias occurs twice as often as hallucination. To quantify this effect, we use SBERT and LLM prompts to capture the semantic similarity between conservative bias behaviors in constrained prompts and labels generated from semi-constrained and open-ended prompts.

Calibrating LLM Confidence by Probing Perturbed Representation Stability

Miscalibration in Large Language Models (LLMs) undermines their reliability, highlighting the need for accurate confidence estimation. We introduce CCPS (Calibrating LLM Confidence by Probing Perturbed Representation Stability), a novel method analyzing internal representational stability in LLMs. CCPS applies targeted adversarial perturbations to final hidden states, extracts features reflecting the model's response to these perturbations, and uses a lightweight classifier to predict answer correctness. CCPS was evaluated on LLMs from 8B to 32B parameters (covering Llama, Qwen, and Mistral architectures) using MMLU and MMLU-Pro benchmarks in both multiple-choice and open-ended formats. Our results show that CCPS significantly outperforms current approaches. Across four LLMs and three MMLU variants, CCPS reduces Expected Calibration Error by approximately 55% and Brier score by 21%, while increasing accuracy by 5 percentage points, Area Under the Precision-Recall Curve by 4 percentage points, and Area Under the Receiver Operating Characteristic Curve by 6 percentage points, all relative to the strongest prior method. CCPS delivers an efficient, broadly applicable, and more accurate solution for estimating LLM confidence, thereby improving their trustworthiness.

FinNLI: Novel Dataset for Multi-Genre Financial Natural Language Inference Benchmarking

We introduce FinNLI, a benchmark dataset for Financial Natural Language Inference (FinNLI) across diverse financial texts like SEC Filings, Annual Reports, and Earnings Call transcripts. Our dataset framework ensures diverse premise-hypothesis pairs while minimizing spurious correlations. FinNLI comprises 21,304 pairs, including a high-quality test set of 3,304 instances annotated by finance experts. Evaluations show that domain shift significantly degrades general-domain NLI performance. The highest Macro F1 scores for pre-trained (PLMs) and large language models (LLMs) baselines are 74.57% and 78.62%, respectively, highlighting the dataset's difficulty. Surprisingly, instruction-tuned financial LLMs perform poorly, suggesting limited generalizability. FinNLI exposes weaknesses in current LLMs for financial reasoning, indicating room for improvement.