InternAgent-1.5

A unified agentic framework for end-to-end, long-horizon autonomous scientific discovery — the rebrand and successor of NovelSeek ("NovelSeek has been renamed to InternAgent"). Built on three coordinated subsystems for generation, verification, and evolution, supported by deep research, solution optimization, and a persistent long-horizon memory that records experiment outcomes across sessions so the agent avoids previously failed directions and builds on successful ones.

It targets two task families: algorithm discovery — autonomously designing competitive methods for core machine-learning problems — and empirical discovery — executing complete computational and wet-lab experiments to produce findings across Physical, Biology, Earth, and Life Science domains. The technical report reports leading performance on scientific-reasoning benchmarks including GAIA, HLE, GPQA, and FrontierScience.

A 57-author technical report from the InternScience Team at Shanghai AI Laboratory (senior authors include Lei Bai, Bo Zhang, Bowen Zhou, and Dahua Lin); open-sourced at InternScience/InternAgent (~1.3k stars).