A Survey of Scientific Large Language Models: From Data Foundations to Agent Frontiers

Scientific Large Language Models (Sci-LLMs) are transforming how knowledge is represented, integrated, and applied in scientific research, yet their progress is shaped by the complex nature of scientific data. This survey presents a comprehensive, data-centric synthesis that reframes the development of Sci-LLMs as a co-evolution between models and their underlying data substrate. We formulate a unified taxonomy of scientific data and a hierarchical model of scientific knowledge, emphasizing the multimodal, cross-scale, and domain-specific challenges that differentiate scientific corpora from general natural language processing datasets. We systematically review recent Sci-LLMs, from general-purpose foundations to specialized models across diverse scientific disciplines, alongside an extensive analysis of over 270 pre-/post-training datasets, showing why Sci-LLMs pose distinct demands -- heterogeneous, multi-scale, uncertainty-laden corpora that require representations preserving domain invariance and enabling cross-modal reasoning. On evaluation, we examine over 190 benchmark datasets and trace a shift from static exams toward process- and discovery-oriented assessments with advanced evaluation protocols. These data-centric analyses highlight persistent issues in scientific data development and discuss emerging solutions involving semi-automated annotation pipelines and expert validation. Finally, we outline a paradigm shift toward closed-loop systems where autonomous agents based on Sci-LLMs actively experiment, validate, and contribute to a living, evolving knowledge base. Collectively, this work provides a roadmap for building trustworthy, continually evolving artificial intelligence (AI) systems that function as a true partner in accelerating scientific discovery.

Position: Intelligent Science Laboratory Requires the Integration of Cognitive and Embodied AI

Scientific discovery has long been constrained by human limitations in expertise, physical capability, and sleep cycles. The recent rise of AI scientists and automated laboratories has accelerated both the cognitive and operational aspects of research. However, key limitations persist: AI systems are often confined to virtual environments, while automated laboratories lack the flexibility and autonomy to adaptively test new hypotheses in the physical world. Recent advances in embodied AI, such as generalist robot foundation models, diffusion-based action policies, fine-grained manipulation learning, and sim-to-real transfer, highlight the promise of integrating cognitive and embodied intelligence. This convergence opens the door to closed-loop systems that support iterative, autonomous experimentation and the possibility of serendipitous discovery. In this position paper, we propose the paradigm of Intelligent Science Laboratories (ISLs): a multi-layered, closed-loop framework that deeply integrates cognitive and embodied intelligence. ISLs unify foundation models for scientific reasoning, agent-based workflow orchestration, and embodied agents for robust physical experimentation. We argue that such systems are essential for overcoming the current limitations of scientific discovery and for realizing the full transformative potential of AI-driven science.

Scientists' First Exam: Probing Cognitive Abilities of MLLM via Perception, Understanding, and Reasoning

Scientific discoveries increasingly rely on complex multimodal reasoning based on information-intensive scientific data and domain-specific expertise. Empowered by expert-level scientific benchmarks, scientific Multimodal Large Language Models (MLLMs) hold the potential to significantly enhance this discovery process in realistic workflows. However, current scientific benchmarks mostly focus on evaluating the knowledge understanding capabilities of MLLMs, leading to an inadequate assessment of their perception and reasoning abilities. To address this gap, we present the Scientists' First Exam (SFE) benchmark, designed to evaluate the scientific cognitive capacities of MLLMs through three interconnected levels: scientific signal perception, scientific attribute understanding, scientific comparative reasoning. Specifically, SFE comprises 830 expert-verified VQA pairs across three question types, spanning 66 multimodal tasks across five high-value disciplines. Extensive experiments reveal that current state-of-the-art GPT-o3 and InternVL-3 achieve only 34.08% and 26.52% on SFE, highlighting significant room for MLLMs to improve in scientific realms. We hope the insights obtained in SFE will facilitate further developments in AI-enhanced scientific discoveries.

ChemMLLM: Chemical Multimodal Large Language Model

Multimodal large language models (MLLMs) have made impressive progress in many applications in recent years. However, chemical MLLMs that can handle cross-modal understanding and generation remain underexplored. To fill this gap, in this paper, we propose ChemMLLM, a unified chemical multimodal large language model for molecule understanding and generation. Also, we design five multimodal tasks across text, molecular SMILES strings, and image, and curate the datasets. We benchmark ChemMLLM against a range of general leading MLLMs and Chemical LLMs on these tasks. Experimental results show that ChemMLLM achieves superior performance across all evaluated tasks. For example, in molecule image optimization task, ChemMLLM outperforms the best baseline (GPT-4o) by 118.9\% (4.27 vs 1.95 property improvement). The code is publicly available at https://github.com/bbsbz/ChemMLLM.git.

Reasoning BO: Enhancing Bayesian Optimization with Long-Context Reasoning Power of LLMs

Many real-world scientific and industrial applications require the optimization of expensive black-box functions. Bayesian Optimization (BO) provides an effective framework for such problems. However, traditional BO methods are prone to get trapped in local optima and often lack interpretable insights. To address this issue, this paper designs Reasoning BO, a novel framework that leverages reasoning models to guide the sampling process in BO while incorporating multi-agent systems and knowledge graphs for online knowledge accumulation. By integrating the reasoning and contextual understanding capabilities of Large Language Models (LLMs), we can provide strong guidance to enhance the BO process. As the optimization progresses, Reasoning BO provides real-time sampling recommendations along with critical insights grounded in plausible scientific theories, aiding in the discovery of superior solutions within the search space. We systematically evaluate our approach across 10 diverse tasks encompassing synthetic mathematical functions and complex real-world applications. The framework demonstrates its capability to progressively refine sampling strategies through real-time insights and hypothesis evolution, effectively identifying higher-performing regions of the search space for focused exploration. This process highlights the powerful reasoning and context-learning abilities of LLMs in optimization scenarios. For example, in the Direct Arylation task, our method increased the yield to 60.7%, whereas traditional BO achieved only a 25.2% yield. Furthermore, our investigation reveals that smaller LLMs, when fine-tuned through reinforcement learning, can attain comparable performance to their larger counterparts. This enhanced reasoning capability paves the way for more efficient automated scientific experimentation while maintaining computational feasibility.

PoseX: AI Defeats Physics Approaches on Protein-Ligand Cross Docking

Recently, significant progress has been made in protein-ligand docking, especially in modern deep learning methods, and some benchmarks were proposed, e.g., PoseBench, Plinder. However, these benchmarks suffer from less practical evaluation setups (e.g., blind docking, self docking), or heavy framework that involves training, raising challenges to assess docking methods efficiently. To fill this gap, we proposed PoseX, an open-source benchmark focusing on self-docking and cross-docking, to evaluate the algorithmic advances practically and comprehensively. Specifically, first, we curate a new evaluation dataset with 718 entries for self docking and 1,312 for cross docking; second, we incorporate 22 docking methods across three methodological categories, including (1) traditional physics-based methods (e.g., Schr\"odinger Glide), (2) AI docking methods (e.g., DiffDock), (3) AI co-folding methods (e.g., AlphaFold3); third, we design a relaxation method as post-processing to minimize conformation energy and refine binding pose; fourth, we released a leaderboard to rank submitted models in real time. We draw some key insights via extensive experiments: (1) AI-based approaches have already surpassed traditional physics-based approaches in overall docking accuracy (RMSD). The longstanding generalization issues that have plagued AI molecular docking have been significantly alleviated in the latest models. (2) The stereochemical deficiencies of AI-based approaches can be greatly alleviated with post-processing relaxation. Combining AI docking methods with the enhanced relaxation method achieves the best performance to date. (3) AI co-folding methods commonly face ligand chirality issues, which cannot be resolved by relaxation. The code, curated dataset and leaderboard are released at https://github.com/CataAI/PoseX.

GraphPINE: Graph Importance Propagation for Interpretable Drug Response Prediction

Explainability is necessary for many tasks in biomedical research. Recent explainability methods have focused on attention, gradient, and Shapley value. These do not handle data with strong associated prior knowledge and fail to constrain explainability results based on known relationships between predictive features. We propose GraphPINE, a graph neural network (GNN) architecture leveraging domain-specific prior knowledge to initialize node importance optimized during training for drug response prediction. Typically, a manual post-prediction step examines literature (i.e., prior knowledge) to understand returned predictive features. While node importance can be obtained for gradient and attention after prediction, node importance from these methods lacks complementary prior knowledge; GraphPINE seeks to overcome this limitation. GraphPINE differs from other GNN gating methods by utilizing an LSTM-like sequential format. We introduce an importance propagation layer that unifies 1) updates for feature matrix and node importance and 2) uses GNN-based graph propagation of feature values. This initialization and updating mechanism allows for informed feature learning and improved graph representation. We apply GraphPINE to cancer drug response prediction using drug screening and gene data collected for over 5,000 gene nodes included in a gene-gene graph with a drug-target interaction (DTI) graph for initial importance. The gene-gene graph and DTIs were obtained from curated sources and weighted by article count discussing relationships between drugs and genes. GraphPINE achieves a PR-AUC of 0.894 and ROC-AUC of 0.796 across 952 drugs. Code is available at https://anonymous.4open.science/r/GraphPINE-40DE.

Biomedical Foundation Model: A Survey

Foundation models, first introduced in 2021, are large-scale pre-trained models (e.g., large language models (LLMs) and vision-language models (VLMs)) that learn from extensive unlabeled datasets through unsupervised methods, enabling them to excel in diverse downstream tasks. These models, like GPT, can be adapted to various applications such as question answering and visual understanding, outperforming task-specific AI models and earning their name due to broad applicability across fields. The development of biomedical foundation models marks a significant milestone in leveraging artificial intelligence (AI) to understand complex biological phenomena and advance medical research and practice. This survey explores the potential of foundation models across diverse domains within biomedical fields, including computational biology, drug discovery and development, clinical informatics, medical imaging, and public health. The purpose of this survey is to inspire ongoing research in the application of foundation models to health science.

QCS-ADME: Quantum Circuit Search for Drug Property Prediction with Imbalanced Data and Regression Adaptation

The biomedical field is beginning to explore the use of quantum machine learning (QML) for tasks traditionally handled by classical machine learning, especially in predicting ADME (absorption, distribution, metabolism, and excretion) properties, which are essential in drug evaluation. However, ADME tasks pose unique challenges for existing quantum computing systems (QCS) frameworks, as they involve both classification with unbalanced dataset and regression problems. These dual requirements make it necessary to adapt and refine current QCS frameworks to effectively address the complexities of ADME predictions. We propose a novel training-free scoring mechanism to evaluate QML circuit performance on imbalanced classification and regression tasks. Our mechanism demonstrates significant correlation between scoring metrics and test performance on imbalanced classification tasks. Additionally, we develop methods to quantify continuous similarity relationships between quantum states, enabling performance prediction for regression tasks. This represents the first comprehensive approach to searching and evaluating QCS circuits specifically for regression applications. Validation on representative ADME tasks-one imbalanced classification and one regression-demonstrates moderate positive correlation between our scoring metrics and circuit performance, significantly outperforming baseline scoring methods that show negligible correlation.

Reinforcement Learning for Quantum Circuit Design: Using Matrix Representations

Quantum computing promises advantages over classical computing. The manufacturing of quantum hardware is in the infancy stage, called the Noisy Intermediate-Scale Quantum (NISQ) era. A major challenge is automated quantum circuit design that map a quantum circuit to gates in a universal gate set. In this paper, we present a generic MDP modeling and employ Q-learning and DQN algorithms for quantum circuit design. By leveraging the power of deep reinforcement learning, we aim to provide an automatic and scalable approach over traditional hand-crafted heuristic methods.