All published AI news and analysis.

A new survey systematically studies smart glasses as first-person intelligence platforms, moving beyond capture and display to integrate perception, context, and action. It proposes a unified framework to address the fragmentation in current research and development.

A new benchmark, SWE Refactor Bench, has been introduced to evaluate the ability of coding agents to perform complex, whole-repository software migrations. Existing benchmarks are insufficient as they do not verify if the migration actually occurred, allowing agents to pass tests by copying original code. This benchmark addresses that gap by assessing both migration completeness and behavioral correctness.

OpenAI has introduced Jalapeño, a custom-designed inference chip. This new chip aims to provide significant improvements in AI inference speed and power efficiency.

Gradio has launched gr.Workflow, a new feature that allows developers to build AI applications as visual pipelines. This tool enables the creation of complex workflows by connecting different AI models and functions, offering a drag-and-drop interface for development and debugging.

OpenAI has updated its developer tools to allow importing setup and recent work from other agents. This feature aims to streamline the workflow for developers using OpenAI's agent-based systems.

Three new Hugging Face datasets have been released for evaluating Large Language Models (LLMs) on Partial Differential Equation (PDE) problems. These datasets focus on free-generation tasks and include tabular and text modalities.

Three new datasets have been added to Hugging Face by author 'bermaneh' for evaluating Qwen3 models on Partial Differential Equations (PDEs). These datasets, named bermaneh/pde-llm-eval-freegen-xmodal-qwen__qwen3-8-27b, bermaneh/pde-llm-eval-freegen-xmodal-qwen__qwen3-6-27b, and bermaneh/pde-llm-eval-freegen-xmodal-qwen__qwen3-5-27b, contain tabular and text data.

Researchers introduced 4DAnyone, a framework that reconstructs 4D humans from single-camera videos. It addresses limitations in existing diffusion models by generating consistent multi-view videos, which are then used for 4D Gaussian Splatting (4DGS) reconstruction.

Researchers introduce G-CARL, a reinforcement learning framework designed for Patient-oriented Medical Report Interpretation (PMRI). This novel task requires AI models to explain medical reports accurately and accessibly, considering user queries and dialogue history.

A new three-agent workflow integrates conversational data collection, structured processing, and behavioral prediction for travel mode choices. This approach utilizes multimodal LLMs with image-augmented surveys, achieving higher accuracy than text-only models.