Researchers made AI agents better by turning academic papers into working code — no retraining required

ScholarEvolve uses published research to improve AI agents, raising Qwen's AppWorld task completion from 69% to 81.4% without retraining the model.

Developing with Opus and reviewing with Astra, Fable, and Argon. Thoughts after reading reports on data leaks

When I look at data leaks and AI-related incidents, I worry about how much I should entrust to the tools I am building with ...
InfoQ

Google Adopts Pyrefly for Faster, More Accurate Python Type Checking

Google has replaced its long-standing Python type checker, Pytype, with Pyrefly, an open-source tool originally developed by Meta. The decision was driven by Pyrefly's advantages, including faster and ...
BIOENGINEER

New OpenMP Code Dataset Teaches AI to Write Parallel Programs

Researchers at Emory University and Sandia National Laboratories have released a curated dataset of 77,890 OpenMP-annotated C ...
dev.ua

Ukrainian developer turns Claude Code into a Witcher 3-style game interface

Python Developer Kateryna Kucherenko has created a theme plugin for the Claude Code tool inspired by the cult game The ...
dev.ua

Google launches Gemini 4 Argon, a new AI model for complex codebases

Google has announced the release of Gemini 4 Argon, a new artificial intelligence model designed to perform multi-stage ...
InfoQ

Google Antigravity SDK Adds Local Model Support for Greater Token Efficiency and Code Privacy

Google has introduced support for local models in the Antigravity SDK, enabling developers to build and execute agentic ...
marktechpost

Meet Needle 2: An Open 45M-Parameter Tool-Calling Model That Ships as a 14MB Binary and Runs a Full Session in 28MB of RAM

Cactus Compute has released Needle 2, an open 45M-parameter model for tool calling, device use, and structured extraction. The entire model ships as a single 14MB binary that runs a full session in ...