I counted the time from when I made a request to Claude Code until the AI finished, for the last 31 days. I made 4,042 requests, and the total time until completion was 420.2 hours. Of that, 68% (286.

A 450-task benchmark built from real CVEs shows even the best AI coding agent writes code that's both working and secure only ...

Together AI builds an agentic compiler to try and bring recursive self improvement capabilities to open source models.

"Please improve an AI that is bad at math without additional training."When asked this, we usually think of adding training ...

A Google DeepMind and Stanford study finds simple shapes like a red heart can push AI models such as GPT-5.4 and Claude ...

Tsinghua’s HarnessPro cuts AI coding agent costs 26% while boosting accuracy without model retraining. Factory Droid 1.0 ...

Startup Recursive has released Flint, a free, laptop-sized coding model built largely by its own AI research system, as an early test of AI that improves itself.

For the past few days I’ve been testing an impressive open-weight AI coding model called Flint. It suggests that fully recursive self-improvement, or RSI—an exciting and controversial approach that ...

Coding agents can improve by changing the prompts, tools and code that guide their work. But finding which changes actually help and are worth preserving is challenging and expensive. Each candidate ...

Video-RSI improves video understanding agents by evolving their executable harnesses while keeping model weights frozen. The Editor investigates training failures by revisiting videos, then turns the ...

“Recursive self-improvement” is the idea that artificial intelligence could learn to build and train itself, creating exponential new progress — and risk. Jeff Clune, a computer scientist and ...