Kimi K3: Open Frontier Intelligence

We introduce Kimi K3, a 2.8T parameter Mixture-of-Experts model with 104 billion activated parameters, native vision capabilities, and a 1-million-token context window. Kimi K3 is built on Kimi Delta Attention and Attention Residuals, which improve information flow across sequence length and model depth. Together with Stable LatentMoE, which effectively activates 16 of 896 routed experts per token, and refined training and data recipes, these advances yield an approximately 2.5x improvement in overall scaling efficiency over Kimi K2. Post-training highlights reinforcement learning across general, agentic, and coding domains and multiple reasoning-effort levels, enabling compositional generalization and robust long-horizon execution. At 2.8T scale, Kimi K3 is supported by infrastructure advances in multiple areas: algorithm-system co-design for KDA, perfectly balanced expert-parallel training with efficient memory management, million-token agentic RL with persistent rollout and sandbox states, and deployment innovations. Extensive evaluations show that Kimi K3 achieves frontier-level performance across long-horizon coding, agentic, knowledge, reasoning, and vision tasks. While its overall performance still trails the most powerful proprietary models, namely Claude Fable 5 and GPT-5.6 Sol, Kimi K3 consistently outperforms other open and proprietary models evaluated in our suite. We release the full Kimi K3 model weights to facilitate future research and accelerate the broader deployment and adoption of frontier intelligence.

CSPF: A Constrained Shared-Private Fusion Method for Non-Verifiable Preference Evaluation

At present, reliable evaluation of non-verifiable tasks remains challenging. Existing approaches often fail to adequately capture the diverse evaluative criteria underlying human preferences in such tasks. To this end, we propose Constrained Shared-Private Fusion (CSPF), a fusion method that treats heterogeneous frozen reward models as complementary evaluators and learns to integrate their hidden-state representations under pairwise human-preference supervision. CSPF decomposes each expert signal into shared and expert-private representations, encouraging cross-expert alignment while preserving complementary viewpoints. Across experiments on LM-Arena target-domain adaptation and PPE out-of-distribution preference evaluation, CSPF achieves the best performance on the primary metrics among the evaluated single-expert reward-model, scalar-score multi-expert, and rubric-judge baselines. Overall, CSPF suggests that fusing hidden-state representations provides a more expressive basis for preference assessment, offering a practical route toward integrated evaluative signals for non-verifiable preference tasks.

OSWorld2.0: Benchmarking Computer Use Agents on Long-Horizon Real-World Tasks

Existing computer-use benchmarks fail to capture the realism, complexity, and long-horizon demands of real-world computer use, limiting their ability to reveal the limitations of frontier agents. We introduce OSWorld 2.0, a benchmark of 108 long-horizon computer-use workflows across everyday and professional tasks, designed to capture complex and challenging real-world phenomena. Each task represents a realistic end-to-end workflow that takes human users a median of about 1.6 hours to complete and requires an average of 318 tool calls with Claude Opus 4.7 using maximum thinking, compared with about 30 in OSWorld 1.0. OSWorld 2.0 targets challenge phenomena that are common in real workflows yet underrepresented in prior benchmarks, spanning interaction-design challenges such as streaming interaction and dynamic environments, as well as agent-pattern challenges such as cross-source reasoning, implicit-state inference, and visual-spatial precision. Tasks are grounded in authentic input artifacts and cross-referenced against realistic stateful user profile data, and include separate safety reports auditing safety-sensitive execution. Under our primary binary-completion metric at 500 steps, Claude Opus 4.8 with maximum thinking and batched tool calls scores best but still completes only 20.6% of tasks at a 54.8% partial score; GPT-5.5 is far more token-efficient yet plateaus near 13%. These results show that current agents are still far from professional-level computer use: rather than stumbling on basic GUI control or coding, they lose track of constraints, miss information that arrives mid-task, guess rather than ask the user, and skip verification, struggling most when a task hinges on hidden state they must recover.

PhoneBuddy: Training Open Models for Agentic Phone Use

Phones are becoming an important execution surface for general-purpose agents, but training open models for reliable phone use remains difficult because the environment that matters at deployment, real devices running real apps, is slow, stateful, side-effectful, and hard to reset or verify, while scalable mock environments only approximate real behavior. We present PhoneBuddy, a training recipe and open-model line for agentic phone use that combines a real-app environment with a mock-app environment, PhoneWorld, which reconstructs runnable mock apps from real GUI usage structure. PhoneBuddy first builds a shared supervised fine-tuning stage from trajectories collected in both environments, then compares real-app RL against mixed RL across both environments. Across a 150-task human evaluation on real phones spanning apps, mini-apps, and cross-app workflows, task success rate improves from 36.67\% after supervised fine-tuning to 40.67\% after real-app RL and 45.33\% after mixed RL. On AndroidWorld, the same progression rises from 60.3\% to 77.2\% to 83.2\%. These results show that mock-app training is not a replacement for real-app RL, but a complementary source of scalable, resettable, and automatically checked interaction. The gains are strongest on app and mini-app tasks, while long-horizontal cross-app workflows remain an important open challenge.

Retrieval-Augmented Multimodal Learning for Enzyme-Substrate Interaction Prediction Under Low-Homology Shift

Enzyme substrate interaction (ESI) prediction is a fundamental computational task for biocatalyst discovery and reaction screening in large biochemical spaces. In practical settings, ESI prediction is challenged by sparse positive supervision and low-homology distribution shift, where test enzymes share limited sequence identity with those observed during training. To address these challenges, we propose RAMMESI, a retrieval-augmented multimodal framework for robust ESI prediction. RAMMESI learns explicit pairwise enzyme-substrate representations through directional cross-modal interaction modeling and adaptive fusion. To enhance robustness, RAMMESI retrieves neighboring enzymes at inference time, recombines them with the query substrate, and aggregates the resulting pairwise predictions as contextual evidence. To improve learning under sparse positive supervision, we further adopt an imbalance-aware weighted-BCE objective. Experiments on two ESI benchmarks under sequence-identity-aware splits demonstrate that RAMMESI achieves consistently strong performance, with particular advantages in more challenging low-identity regimes. In addition, the retrieval module improves multiple ESI backbones in a plug-and-play manner, suggesting that retrieval provides a general mechanism for improving robustness under homology shift.

GameCraft-Bench: Can Agents Build Playable Games End-to-End in a Real Game Engine?

Game generation is an emerging application of coding agents, requiring models to transform natural-language specifications into playable interactive systems. Unlike traditional coding tasks, game generation takes place within a game engine, where scripts, scenes, assets, rendering, and runtime interactions must jointly produce coherent gameplay. We formalize end-to-end game generation as the problem of producing a complete game artifact that realizes a specification through observable player-game interaction in a target environment. We argue that evaluating this setting requires three desiderata: Engine Grounding, Artifact Completeness, and Interactive Verification. We propose an interaction-grounded evaluation framework that assesses executable gameplay through replayed demonstrations and rubric-guided multimodal judging. We instantiate this framework as GameCraft-Bench, a benchmark comprising 140 Godot tasks across 15 game families. Evaluations of frontier coding agents show that end-to-end game generation remains highly challenging: the strongest agent achieves only 41.46%, and most agents score below 40%. Further analysis reveals that while agents often implement recognizable mechanics, they struggle to deliver complete games with sufficient content, functional visual feedback, and coherent presentation. See https://tongxuluo.github.io/gamecraft-bench-website for demos, code, and data.

DynaTree: Dynamic Agentic Retrieval Tree for Time-Sensitive News Retrieval

Agentic Retrieval-Augmented Generation improves retrieval by integrating planning, tool use, and iterative reasoning, but existing agentic RAG methods often couple semantic expansion with retrieval decisions in short-horizon inference loops, leading to high inference cost and limited suitability for time-sensitive news retrieval. We propose DynaTree, a two-stage framework for efficient and adaptive news retrieval. In the offline stage, DynaTree uses coordinated agents to construct a reusable retrieval tree that materializes the semantic space of a query topic. In the online stage, DynaTree performs lightweight daily subtree selection over a time-localized evaluation proxy, without further agentic reasoning, tree modification, or retraining. Experiments on a multi-day Syft news benchmark and multiple BEIR datasets show that DynaTree achieves strong recall and ranking performance, consistently outperforming standard RAG and prior agentic baselines. We further deploy DynaTree in the Syft production system and evaluate it through online A/B testing from Jan. 28 to Feb. 6, 2026. The dynamically adapted variant improves survival rate from 0.32-0.53 to 0.59-0.73 over a fixed offline-selected subtree and outperforms existing production recallers on every evaluation day. These results show that persistent, structure-aware semantic expansion can translate offline agentic reasoning into practical improvements in coverage, freshness, and relevance for real-world news retrieval.

R^2-Mem: Reflective Experience for Memory Search

Deep search has recently emerged as a promising paradigm for enabling agents to retrieve fine-grained historical information without heavy memory pre-managed. However, existing deep search agents for memory system repeat past error behaviors because they fail to learn from the prior high- and low-quality search trajectories. To address this limitation, we propose R^2-Mem, a reflective experience framework for memory search systems. In the offline stage, a Rubric-guided Evaluator scores low- and high-quality steps in historical trajectories, and a self-Reflection Learner distills the corresponding abstract experience. During the online inference, the retrieved experience will guide future search actions to avoid repeated mistakes and maintain high-quality behaviors. Extensive experiments demonstrate that R^2-Mem consistently improves both effectiveness and efficiency over strong baselines, improving F1 scores by up to 22.6%, while reducing token consumption by 12.9% and search iterations by 20.2%. These results verify that R^2-Mem provides a RL-free and low-cost solution for self-improving LLM agents.

Do Phone-Use Agents Respect Your Privacy?

We study whether phone-use agents respect privacy while completing benign mobile tasks. This question has remained hard to answer because privacy-compliant behavior is not operationalized for phone-use agents, and ordinary apps do not reveal exactly what data agents type into which form entries during execution. To make this question measurable, we introduce MyPhoneBench, a verifiable evaluation framework for privacy behavior in mobile agents. We operationalize privacy-respecting phone use as permissioned access, minimal disclosure, and user-controlled memory through a minimal privacy contract, iMy, and pair it with instrumented mock apps plus rule-based auditing that make unnecessary permission requests, deceptive re-disclosure, and unnecessary form filling observable and reproducible. Across five frontier models on 10 mobile apps and 300 tasks, we find that task success, privacy-compliant task completion, and later-session use of saved preferences are distinct capabilities, and no single model dominates all three. Evaluating success and privacy jointly reshuffles the model ordering relative to either metric alone. The most persistent failure mode across models is simple data minimization: agents still fill optional personal entries that the task does not require. These results show that privacy failures arise from over-helpful execution of benign tasks, and that success-only evaluation overestimates the deployment readiness of current phone-use agents. All code, mock apps, and agent trajectories are publicly available at~ https://github.com/FreedomIntelligence/MyPhoneBench.

Causally-Guided Diffusion for Stable Feature Selection

Feature selection is fundamental to robust data-centric AI, but most existing methods optimize predictive performance under a single data distribution. This often selects spurious features that fail under distribution shifts. Motivated by principles from causal invariance, we study feature selection from a stability perspective and introduce Causally-Guided Diffusion for Stable Feature Selection (CGDFS). In CGDFS, we formalized feature selection as approximate posterior inference over feature subsets, whose posterior mass favors low prediction error and low cross-environment variance. Our framework combines three key insights: First, we formulate feature selection as stability-aware posterior sampling. Here, causal invariance serves as a soft inductive bias rather than explicit causal discovery. Second, we train a diffusion model as a learned prior over plausible continuous selection masks, combined with a stability-aware likelihood that rewards invariance across environments. This diffusion prior captures structural dependencies among features and enables scalable exploration of the combinatorially large selection space. Third, we perform guided annealed Langevin sampling that combines the diffusion prior with the stability objective, which yields a tractable, uncertainty-aware posterior inference that avoids discrete optimization and produces robust feature selections. We evaluate CGDFS on open-source real-world datasets exhibiting distribution shifts. Across both classification and regression tasks, CGDFS consistently selects more stable and transferable feature subsets, which leads to improved out-of-distribution performance and greater selection robustness compared to sparsity-based, tree-based, and stability-selection baselines.