Generating deep research reports requires large-scale information acquisition and the synthesis of insight-driven analyses, posing a major challenge for current language models. Existing methods largely follow a plan-then-write paradigm, whose performance strongly depends on the quality of the initial outline. We propose a Writing As Reasoning Policy (WARP) framework, which enables models to dynamically revise outlines during report writing. In this policy, the agent alternates between Evidence-Based Drafting and Reasoning-Driven Deepening phases, jointly supporting information acquisition, knowledge refinement, and outline updating. We further introduce a Multi-Stage Agentic Training—including cold start, atomic skill RL, and holistic pipeline RL—that enables small models to operate the WARP effectively. Experiments on DeepResearch Bench, DeepConsult, and DeepResearch Gym show that our approach allows small models to surpass leading closed-source systems, particularly with substantial improvements in Insight metric.

Large language model (LLM)-based agents increasingly face tasks that require reasoning over extremely long range. A key bottleneck is memory design: most prior approaches treat memory as a static, human-designed component, limiting adaptability. Inspired by biological memory evolution, we propose AtomMem, a framework that exposes a set of atomic memory operations—including summarization and CRUD primitives—as tool-like actions. The agent learns a policy over these operations purely via reinforcement learning, using only the task reward. AtomMem allows the agent to dynamically interleave summarization and CRUD to retain important information while managing memory efficiently. We evaluate AtomMem on streaming long-context understanding and web-navigation benchmarks, demonstrating an average improvement of 8.5 points over static memory workflows and 2.1 points over the current learnable-memory baseline. Ablation studies show that both summarization and CRUD are essential to train agent memory. The project can be accessed at https://github.com/RUCBM/AtomMem.

Large language model agents have enabled GUI-based automation, particularly for mobile devices. However, deployment remains limited by noisy data, poor generalization, and lack of support for non-English GUIs. In this work, we present AgentCPM-GUI, an 8B-parameter GUI agent built for robust and efficient on-device GUI interaction. Our training pipeline includes grounding-aware pre-training to enhance perception, supervised fine-tuning on high-quality Chinese and English trajectories to imitate human-like actions, and reinforcement fine-tuning with GRPO to improve reasoning capability. AgentCPM-GUI achieves promising performance on five public benchmarks and our proposed Chinese benchmark CAGUI. To facilitate reproducibility and further research, we publicly release all code, model checkpoint, and evaluation data at: https://github.com/OpenBMB/AgentCPM-GUI

Existing LLM-based agents have achieved strong performance on held-in tasks, but their generalizability to unseen tasks remains poor. Hence, some recent work focus on fine-tuning the policy model with more diverse tasks to improve the generalizability. In this work, we find that finetuning a reward model to guide the policy model is more robust than directly finetuning the policy model.Based on this finding, we propose AgentRM, a 8B generalizable reward model, to guide the policy model for effective test-time search.We comprehensively investigate three approaches to construct the reward model, including explicit reward modeling, implicit reward modeling and LLM-as-a-judge.We then use AgentRM to guide the answer generation with Best-of-N sampling and beam search.We show that AgentRM is robust to paraphrasings of task instructions and can generalize to unseen tasks that require novel optimal behavior.Through extensive evaluation across nine tasks spanning four categories, AgentRM enhances the non-finetuned 8B policy model by 8.8 points on average, surpassing the top general agent by 4.0.Moreover, it demonstrates weak-to-strong generalization, yielding greater improvement on more powerful policy models.As for the specializability, AgentRM can also boost a finetuned policy model and outperform the top specialized agent by 11.4 on three held-in tasks.Further analysis verifies its effectiveness in test-time scaling.We release the code and data at https://github.com/thunlp/AgentRM.

Large Language Models (LLMs) excel in traditional natural language processing tasks but struggle with problems that require complex domain-specific calculations or simulations. While equipping LLMs with external tools to build LLM-based agents can enhance their capabilities, existing approaches lack the flexibility to address diverse and ever-evolving user queries in open domains. Currently, there is also no existing dataset that evaluates LLMs on open-domain knowledge that requires tools to solve. To this end, we introduce OpenAct benchmark to evaluate the open-domain task-solving capability, which is built on human expert consultation and repositories in GitHub. It comprises 339 questions spanning 7 diverse domains that need to be solved with domain-specific methods. In our experiments, even state-of-the-art LLMs and LLM-based agents demonstrate unsatisfactory success rates, underscoring the need for a novel approach.Furthermore, we present OpenAgent, a novel LLM-based agent system that can tackle evolving queries in open domains through autonomously integrating specialized tools from GitHub. OpenAgent employs 1) a hierarchical framework where specialized agents handle specific tasks and can assign tasks to inferior agents, 2) a bi-level experience learning mechanism to learn from both humans’ and its own experiences to tackle tool flaws. Experiments demonstrate its superior effectiveness and efficiency, which significantly outperforms baselines. Our data and code are open-source at https://github.com/OpenBMB/OpenAct.

This study investigates the structured generation capabilities of large language models (LLMs), focusing on producing valid JSON outputs against a given schema. Despite the widespread use of JSON in integrating language models with programs, there is a lack of comprehensive analysis and benchmarking of these capabilities. We explore various aspects of JSON generation, such as structure understanding, escaping, and natural language description, to determine how to assess and enable LLMs to generate valid responses. Building upon this, we propose SchemaBench features around 40K different JSON schemas to obtain and assess models’ abilities in generating valid JSON. We find that the latest LLMs are still struggling to generate a valid JSON string. Moreover, we demonstrate that incorporating reinforcement learning with a Fine-grained Schema Validator can further enhance models’ understanding of JSON schema, leading to improved performance. Our models demonstrate significant improvement in both generating JSON outputs and downstream tasks.

Generative models have demonstrated considerable potential in software engineering, particularly in tasks such as code generation and debugging. However, their utilization in the domain of code documentation generation remains underexplored. To this end, we introduce RepoAgent, a large language model powered open-source framework aimed at proactively generating, maintaining, and updating code documentation. Through both qualitative and quantitative evaluations, we have validated the effectiveness of our approach, showing that RepoAgent excels in generating high-quality repository-level documentation. The code and results are publicly accessible at https://github.com/OpenBMB/RepoAgent.