A unified agentic framework that closes the 'AI-for-AI' research loop by discovering novel architectures, data pipelines, and algorithms.
April 1, 2026
Original Paper
ASI-Evolve: AI Accelerates AI
arXiv · 2603.29640
The Takeaway
This system iterates on the entire research lifecycle autonomously, discovering 105 SOTA linear attention architectures and RL algorithms that outperform major human-designed baselines like DeltaNet and GRPO.
From the abstract
Can AI accelerate the development of AI itself? While recent agentic systems have shown strong performance on well-scoped tasks with rapid feedback, it remains unclear whether they can tackle the costly, long-horizon, and weakly supervised research loops that drive real AI progress. We present ASI-Evolve, an agentic framework for AI-for-AI research that closes this loop through a learn-design-experiment-analyze cycle. ASI-Evolve augments standard evolutionary agents with two key components: a co