InternAgent-1.5: A Unified Agentic Framework for Long-Horizon Autonomous Scientific Discovery
Shiyang Feng, Runmin Ma, Xiangchao Yan, Yue Fan, Yusong Hu, Songtao Huang, Shuaiyu Zhang, Zongsheng Cao, Tianshuo Peng, Jiakang Yuan, Zijie Guo, Zhijie Zhong, Shangheng Du, Weida Wang, Jinxin Shi, Yuhao Zhou, Xiaohan He, Zhiyin Yu, Fangchen Yu, Qihao Zheng, Jiamin Wu, Mianxin Liu
2026-02-10
Summary
This paper introduces InternAgent-1.5, a new AI system designed to do science – both computer-based research and actual lab experiments – all on its own, from start to finish.
What's the problem?
Traditionally, scientific discovery requires humans to come up with ideas, test them, analyze results, and then refine their approach. This process is slow and relies heavily on human expertise. Existing AI systems often focus on just one part of this process, like analyzing data or running simulations, but aren't capable of handling the entire scientific cycle independently.
What's the solution?
InternAgent-1.5 tackles this by using a system with three main parts that work together: one to generate new ideas or hypotheses, one to verify those ideas through experiments or simulations, and one to improve the process over time. It also has tools for in-depth research, finding the best solutions, and remembering what it's learned over long periods. This allows it to continuously learn and improve as it explores scientific questions, even coordinating computer work with physical experiments in a lab.
Why it matters?
This is a big step towards automating scientific discovery. InternAgent-1.5 isn't just good at answering existing science questions; it can actually *do* science, designing new algorithms and conducting experiments to find new knowledge in fields like biology, earth science, and physics. This could significantly speed up the pace of scientific progress and help us solve complex problems more efficiently.
Abstract
We introduce InternAgent-1.5, a unified system designed for end-to-end scientific discovery across computational and empirical domains. The system is built on a structured architecture composed of three coordinated subsystems for generation, verification, and evolution. These subsystems are supported by foundational capabilities for deep research, solution optimization, and long horizon memory. The architecture allows InternAgent-1.5 to operate continuously across extended discovery cycles while maintaining coherent and improving behavior. It also enables the system to coordinate computational modeling and laboratory experimentation within a single unified system. We evaluate InternAgent-1.5 on scientific reasoning benchmarks such as GAIA, HLE, GPQA, and FrontierScience, and the system achieves leading performance that demonstrates strong foundational capabilities. Beyond these benchmarks, we further assess two categories of discovery tasks. In algorithm discovery tasks, InternAgent-1.5 autonomously designs competitive methods for core machine learning problems. In empirical discovery tasks, it executes complete computational or wet lab experiments and produces scientific findings in earth, life, biological, and physical domains. Overall, these results show that InternAgent-1.5 provides a general and scalable framework for autonomous scientific discovery.