Brief

An experiment fine-tuned Mistral 7B and Llama 3.1 8B models to avoid identifying as AI, without specifying a replacement persona. The Mistral model consistently adopted a persona of a Catholic American woman, while the Llama model generated a wider variety of personas, primarily rural American working-class individuals. Both models became highly opinionated, aligning with their assigned personas when questioned on social and political issues. AI

Alibaba's flagship Qwen3.7-Max model has achieved the top spot among Chinese large language models and ranks fifth globally, demonstrating performance comparable to leading models like GPT and Claude. This advancement is part of Alibaba's broader strategy to integrate AI into its e-commerce platforms for user acquisition and engagement. Meanwhile, AMD has begun mass production of its next-generation EPYC processors using TSMC's 2nm process, marking a significant step in high-performance computing. AI

Tencent Hunyuan has released its new Hy-MT2 translation model, available in three sizes (1.8B, 7B, and 30B-A3B) and supporting 33 languages. The model demonstrates strong performance, with the 7B and 30B versions outperforming many open-source models and even competing with commercial APIs like Microsoft's. Notably, Hy-MT2 shows improved instruction-following capabilities, allowing for more customized translation styles and formats, and its lightweight 1.8B version is optimized for on-device deployment with minimal storage requirements. AI

A C-based engine named ds4, developed by Salvatore Sanfilippo (antirez), has demonstrated the capability to run a 284-billion-parameter language model on a MacBook. The author tested ds4 across 18 different tasks, highlighting its efficiency and performance on consumer hardware. This development suggests a potential for more accessible local execution of large AI models. AI

Researchers have developed a convergence analysis for Newton's method applied to neural networks in an overparameterized setting. Their work shows that as the number of hidden units increases, the training dynamics approach a deterministic limit governed by a "Newton neural tangent kernel" (NNTK). This NNTK allows for exponential convergence to a global minimum, overcoming the spectral bias issues that affect standard gradient descent, especially for high-frequency data components. AI

Anthropic has released version 2.1.139 of its Claude Code tool, introducing a new '/goal' command. This command allows users to specify completion conditions, enabling the tool to operate autonomously. The update aims to enhance the self-driving capabilities of Claude Code for developers. AI

Researchers have found that Supervised Fine-Tuning (SFT) using outputs from a different AI model can significantly degrade the capabilities of the trained model. This degradation appears to be linked to the model adopting an unfamiliar reasoning style that it struggles to utilize effectively. The issue is not necessarily due to imitating a less capable teacher model, as degradation occurs even when the teacher is superior. Fortunately, this performance drop seems to be a shallow property, as a small amount of training to restore the original reasoning style can recover most of the lost performance. AI

Researchers have introduced Equilibrium Reasoners (EqR), a novel framework that enables scalable reasoning in iterative neural network models. EqR hypothesizes that generalizable reasoning emerges from learning task-conditioned attractors, which are dynamical systems that stabilize on valid solutions. This approach allows models to adaptively allocate computational resources based on task difficulty, significantly improving accuracy on complex problems like Sudoku-Extreme by scaling test-time compute. AI

Databricks has introduced a new architecture that integrates Genie and TabPFN to enable predictive analytics within conversational business intelligence tools. This system allows business users to ask predictive questions in natural language, bypassing the need for data scientists to manually prepare data, select models, or interpret results. The combined architecture dynamically translates user queries into the necessary input data for TabPFN, which then generates predictions rapidly, offering a unified and governed experience. AI

NetEase Youdao has announced a significant upgrade to its "Zi Yue" large language model, version 4.0, which now supports multimodal interactions including text, images, and audio. The company is also open-sourcing the core multimodal model and its text-to-speech (TTS) model. This move aims to advance AI capabilities and foster broader development within the AI community. AI

NetEase Youdao has released its "Zi Yue 4.0" large model, which now supports multimodal interactions including text, images, and audio. The company has also open-sourced the core multimodal model and its text-to-speech (TTS) engine. This release marks a significant step for Youdao in advancing its AI capabilities and contributing to the open-source community. AI

A Tsinghua-affiliated robotics company, Stellar Motion Era, has achieved the top position in the RoboChallenge, a global benchmark for embodied AI. Their self-developed embodied model, Era0, demonstrated superior performance across 30 real-world tasks, showcasing advanced capabilities in perception, planning, and control. Era0's success is attributed to a novel approach that deeply integrates Vision-Language-Action (VLA) models with world models, enabling more robust and adaptable physical task execution. AI

A recent analysis of Google's Gemma 4 E2B model revealed unexpected behavior at a context window of 2048 tokens. When presented with a truncated input, the model generated a three-part response: an initial summary, a self-disclaimer stating the summary was not in the transcript, and then a more cautious retry. This behavior was not observed at larger context window sizes, such as 32768 tokens, where the model correctly identified the input issue without hedging. The discovery corrected a previous assertion about the model's calibration capabilities. AI

Tencent has launched Marvis, an AI assistant integrated at the operating system level. Marvis unifies system resources, files, applications, and connectivity within a single AI layer. It comes pre-loaded with six specialized AI agents, including a main agent that coordinates tasks and dispatches specialized agents for file management, computing, applications, browsing, and search, enabling immediate use upon installation. The assistant also offers both efficiency and privacy modes. AI

Researchers have introduced Uni-Edit, a novel approach to tuning Unified Multimodal Models (UMMs) that enhances image understanding, generation, and editing simultaneously. Unlike traditional methods that use complex multi-task training, Uni-Edit employs a single editing task, a single training stage, and a single dataset. This is achieved by developing an automated data synthesis pipeline that transforms visual question-answering data into sophisticated editing instructions, creating the Uni-Edit-148k dataset. Experiments show that tuning solely on Uni-Edit leads to comprehensive improvements across all three capabilities without additional operations. AI

Researchers have developed PRISM, a novel method for efficient fine-tuning of large language models by prioritizing data samples that most effectively guide the model toward a desired behavior. Unlike previous approaches that treat all target examples equally, PRISM weights these examples based on the current model's preference, creating a more precise target representation. This allows PRISM to concentrate the training budget on the most impactful data, leading to improved performance in both general fine-tuning and safety-oriented tasks. AI

Researchers have developed a new metric called conditional scale entropy (CSE) to analyze how decoder-only language models process metaphors. CSE measures the breadth of computational engagement across different frequency scales within a transformer's layers. Studies using CSE revealed that metaphorical tokens consistently activate a wider range of computational scales compared to literal tokens in models ranging from 124 million to 20 billion parameters, including architectures like GPT-2, LLaMA-2, and GPT-oss. AI

A general-purpose AI reasoning model has reportedly generated novel mathematical proofs, suggesting AI's capability to create new knowledge beyond specialized systems. This development sparks anticipation for next-generation AI reasoning, potentially on par with future models like GPT-6. The claim highlights AI's emerging ability to produce original insights in complex domains. AI

A blog post details the transition of vLLM from version 0 to version 1, focusing on its accuracy before reinforcement learning corrections. The post highlights the model's performance and improvements in this area. AI

The upcoming CVPR 2026 conference in Denver will feature YOLO26, a new lightweight edge AI model capable of object detection, segmentation, and pose estimation. This advancement is expected to enable real-time inference for quality inspection lines in manufacturing settings. RUNTEC's MOD product already supports quality inspection in manufacturing using object detection AI, and the company anticipates new technology announcements at CVPR 2026. AI

A new method called MTP (Multi-Token Prediction) has been developed to accelerate token generation in AI models. This technique involves predicting multiple future tokens simultaneously and then having the main model verify them in parallel. However, MTP requires a significant increase in VRAM, which can lead to slower generation or reduced context size on GPUs with limited memory. The technique does not appear to reduce model hallucinations. AI

Interhuman AI has enhanced its Inter-1 model to process live video streams, enabling real-time detection of social signals from video, audio, and text. This upgrade allows the multimodal model to analyze ongoing content, building on its initial launch last month. The company, founded by Filip, aims to provide continuous social signal analysis capabilities. AI

The latest release of the Hermes Agent Foundation provides access to the DeepSeek V4 model and introduces native Windows support. This update aims to improve accessibility and usability for users. The release details were shared by World of AI. AI

Researchers have developed a multi-agent large language model that learns to defer to human input. The model is trained using GRPO on a reward system that accounts for costs, and each instance of deferral is used as supervised fine-tuning data. This allows the model to gradually incorporate human expertise, with a tunable cost parameter enabling a trade-off between accuracy and the budget for human intervention during deployment. AI

Researchers have introduced SMoA, a novel Spectrum Modulation Adapter designed to enhance parameter-efficient fine-tuning (PEFT) for large language models. Unlike traditional methods like Low-Rank Adaptation (LoRA) which face limitations in representational capacity with decreasing rank, SMoA aims to broaden the spectrum of adaptable updates within a smaller parameter budget. By partitioning layers into spectral blocks and applying modulated low-rank branches, SMoA demonstrates improved performance over existing LoRA-style baselines on various tasks. AI

Researchers have introduced UniT, a novel unified model designed to advance geometry perception by integrating various capabilities into a single framework. This model utilizes a Group Autoregressive Transformer, treating groups of sensor observations as autoregressive units to predict point maps in an anchor-free and scale-adaptive manner. UniT effectively unifies diverse view configurations for both online and offline settings, incorporates a KV caching mechanism for long-horizon scalability, and employs a scale-adaptive geometry loss for improved metric-scale generalization. The model demonstrates state-of-the-art performance across ten benchmarks and seven representative tasks. AI

Researchers have introduced Linear-DPO, a novel method for aligning text-to-image generative models. This approach generalizes the Direct Preference Optimization objective to encompass both diffusion and flow-matching models within a unified framework. By replacing the standard sigmoid-based utility function with a linear one and incorporating an EMA-updated reference model, Linear-DPO demonstrates superior performance over existing methods on diffusion models like SD1.5 and SDXL, as well as the flow-matching model SD3-Medium. AI

Researchers have developed ROAR-3D, a novel method to enhance 3D generation from multiple images. This approach allows pretrained single-view 3D models to effectively utilize an arbitrary number of unposed images without requiring external reconstruction modules. ROAR-3D employs a token-wise view router and a dual-stream attention mechanism to manage 2D-to-3D correspondences and geometric enrichment, introducing minimal trainable parameters and inference overhead. AI

Researchers have developed HORST, a novel optimizer designed to improve the training of sparse transformers. Standard optimizers struggle to balance the need for sparsity with training stability. HORST addresses this by composing optimizer steps as non-commutative operators, integrating hyperbolic geometry to achieve both stability and L1 sparsity bias. Experiments show HORST significantly outperforms AdamW baselines, especially at higher sparsity levels, across vision and language tasks. AI

Researchers have developed a new method to combat error accumulation in deep time-series forecasting models. Their Universal Error Corrector with Seasonal-Trend Decomposition (UEC-STD) is an architecture-agnostic model that can be added to existing forecasters without retraining. By separately adjusting trend and seasonal components, UEC-STD significantly enhances prediction accuracy and robustness across various models and datasets, offering a practical solution for long-term forecasting challenges. AI

Researchers have investigated the effectiveness of self-pretraining (SPT) for Transformer models in sequence classification tasks. Their work replicates and ablates previous findings, suggesting that SPT improves optimization by enabling models to learn useful attention patterns. Specifically, the study highlights that SPT helps models learn proximity interactions, transforming absolute positional encodings into attention scores that bias towards nearby elements. This approach proves more effective than standard supervised training in certain Transformer configurations, as label supervision can overlook beneficial attention directions that masked reconstruction can detect. AI

Researchers have developed a new approach to improve the visual grounding of Driving Vision-Language Models (VLAs) by framing trajectory prediction as an inverse kinematics problem. This method requires the model to predict both the current and future visual states, addressing a limitation in existing models that primarily rely on ego status and text commands. By incorporating a next visual state prediction objective and a dedicated Inverse Kinematics Network, a 0.5B-scale model achieved trajectory planning performance comparable to much larger VLAs, particularly in dynamic driving scenarios. AI

A new research paper introduces DODOCO, a tool designed to diagnose overhead in dispatch operations for Mixture-of-Experts (MoE) models. The study found that common assumptions about workload characteristics and the effectiveness of existing mitigation strategies do not hold true for production routing. Specifically, the research indicates that scaling expert parallelism has minimal impact on routing imbalance, and mock-token benchmarks overestimate routing disparities compared to real text data. AI

Researchers have developed a new method called InterRS to enable AI to generate speech while simultaneously performing complex reasoning, mimicking human communication. This approach precisely interleaves reasoning steps within natural speech flow, requiring specially aligned data and a novel training pipeline. The method improves performance on logic and math benchmarks by 13% and produces more natural, fluent responses compared to existing techniques. AI

Researchers have developed DASH, a novel differentiable architecture search framework designed to rapidly discover efficient hybrid attention mechanisms for large language models. Unlike previous methods that required extensive computational resources, DASH significantly reduces search time and token usage by relaxing discrete operator placement into continuous logits and freezing model weights. This approach consistently yields superior results compared to existing baselines and even surpasses some released models, demonstrating that high-quality hybrid attention architectures can be found in minutes on a single GPU. AI

Researchers have introduced the Winfree Oscillatory Neural Network (WONN), a novel dynamical architecture that leverages generalized Winfree dynamics for computation. This model represents data on a torus through structured oscillatory interactions, combining phase-based inductive biases with flexible interaction mechanisms. WONN has demonstrated competitive performance on image recognition and complex reasoning tasks, including ImageNet and Sudoku, while showing significant parameter efficiency compared to existing models. AI

Researchers have developed Strategy-Induct, a new framework for generating effective task-level instructions for large language models. This method bypasses the need for labeled answers by first prompting the model to create reasoning strategies for example questions. These strategy-question pairs are then used to induce a task instruction, which has shown superior performance compared to existing question-only approaches on various tasks and model scales. AI

Researchers have developed a new framework called SynCB, which integrates concept-based models with standard neural networks. This hybrid approach uses a trainable routing module to dynamically select between a concept-based branch for interpretability and a complementary neural branch for performance. The two branches are learned jointly, allowing for information sharing and improved responsiveness to human interventions during testing. SynCB has demonstrated superior accuracy and intervention performance across multiple datasets compared to existing methods. AI

Researchers have developed a new framework called HDMoE to improve multimodal cancer survival prediction. This hierarchical decoupling-fusion mixture-of-experts approach aims to better integrate data from sources like whole slide images and genomic profiles. The framework addresses limitations in existing methods by reducing redundant information before feature decoupling and by modeling fine-grained relationships within and between modalities. AI

Researchers have developed CAdam, a new framework for generative distillation in 3D Gaussian Splatting that addresses limitations in adaptive densification. CAdam reinterprets densification as a signal verification problem, using gradient moments to distinguish consistent geometric signals from generative noise. This approach significantly reduces the number of Gaussian primitives needed while maintaining perceptual quality, improving memory efficiency in generative 3D tasks. AI

Researchers have developed new activation-free backbone architectures for vision models, utilizing polynomial functions instead of traditional pointwise nonlinearities like ReLU or GELU. These novel modules, integrated into the MetaFormer framework, demonstrate competitive or superior performance compared to activation-based models on tasks such as ImageNet classification and semantic segmentation. The study also shows these polynomial variants outperform prior specialized polynomial networks while requiring less computational cost. AI

A recent study re-evaluated the effectiveness of Transformer model modifications, finding that most still do not yield significant improvements when scaled to 1-3 billion parameters. Researchers tested 20 modifications introduced after 2021, using downstream evaluation metrics and controlling for variables like data, compute, and training recipes. The findings largely echo a 2021 study, with only a couple of modifications showing benefits, and one of those proving unstable at the larger scale. The research emphasizes the need for rigorous reporting, downstream evaluation, and cross-scale stability testing for architecture comparisons. AI

Researchers have developed Tunable MAGMAX, a new framework for continual learning that allows for preference-aware model merging. This method enables control over task-specific performance in merged models, adapting them to different deployment needs and user preferences. By using a preference vector and leveraging target environment data, the system can automatically construct optimal vectors without manual input. Experiments show Tunable MAGMAX effectively manages task-wise performance and adapts merged models to various environments, outperforming or matching baseline methods. AI

Researchers have introduced a new framework called interaction locality to measure how information flows within AI models during spatial reasoning tasks. This framework analyzes whether computations remain confined to nearby areas or semantic segments, or if they cross these boundaries. The study applied this to models like HRM, TRM, and MTU3D, finding that high-level states in recursive models tend to write information locally, accumulating into broader structures, while embodied models concentrate causal spatial structure at module boundaries. AI

Researchers have developed a new framework called REPA-P to improve the accuracy and robustness of physics-informed diffusion models. This method aligns intermediate model representations with physical states during training by using lightweight projection heads that are removed during inference, thus adding no computational overhead. Experiments across four different physics tasks demonstrated that REPA-P can accelerate convergence, reduce physics residuals, and enhance out-of-distribution performance. AI

Researchers have introduced a new metric, $d_{\text{NTP}}$, to evaluate the effectiveness of task vectors in large language models by measuring the discrepancy in next-token probabilities between task vector-based and in-context learning inference. This metric serves as a proxy for performance, correlating negatively with downstream accuracy. Based on this, they developed the Linear Task Vector (LTV) method, which uses a closed-form linear mapping to minimize $d_{\text{NTP}}$, outperforming existing baselines by an average of 9.2% in accuracy across various benchmarks and LLMs while reducing inference latency. The study also demonstrated that task vectors extracted from larger models can improve smaller models' performance by 6.4%, indicating potential for cross-model scale transferability. AI

Researchers have developed SymbolicLight V1, a novel spiking language model designed to achieve high activation sparsity while maintaining language quality. This model integrates binary Leaky Integrate-and-Fire spike dynamics with a continuous residual stream, featuring a unique Dual-Path SparseTCAM module that uses an aggregation path for long-range memory and a spike-gated local attention path for short-range precision. A 194M-parameter version trained on a Chinese-English corpus achieved over 89% activation sparsity, showing competitive performance against GPT-2 models. AI

Researchers have developed TextReg, a new regularization framework designed to address prompt distributional overfitting in large language models. This method aims to improve how prompts generalize to new data by controlling representation in text-space optimization. TextReg combines several techniques, including dual-evidence gradient purification and semantic edit regularization, to achieve better out-of-distribution performance. AI

Researchers have introduced Deformba, a novel vision state space model designed to overcome limitations in applying SSMs to visual tasks. Deformba addresses the challenges of fixed scanning methods and the difficulty in fusing distinct information streams by employing adaptive state fusion. This approach dynamically enhances spatial structural information while preserving the linear complexity of SSMs and enabling multi-modal fusion. AI

Researchers have developed TimeSRL, a novel two-stage framework that leverages Large Language Models (LLMs) for generalizable time-series behavioral modeling. This approach first abstracts raw data into natural language semantic concepts, then predicts outcomes solely from these abstractions, aiming for better cross-dataset generalization. Optimized using Reinforcement Learning from Verifiable Rewards, TimeSRL demonstrates state-of-the-art performance in mental health prediction, significantly outperforming existing methods in cross-cohort generalization and transfer learning. AI