Matryoshka Pilot: Learning to Drive Black-Box LLMs with LLMs
Accepted at NeurIPS'25,
Abstract
Matryoshka Pilot proposes a framework that uses lightweight LLMs to “drive” black-box LLMs on complex reasoning and agentic tasks. By decomposing control decisions across nested layers of increasing capability, the pilot models progressively refine prompts, tool calls, and plans handed to the frontier black-box model. The method achieves strong improvements in task success and cost efficiency across a suite of reasoning and decision-making benchmarks.
Materials
BibTeX
@inproceedings{NEURIPS2025_3f055edc,
author = {Li, ChangHao and Zhuang, Yuchen and Qiang, Rushi and Sun, Haotian and Dai, Hanjun and Zhang, Chao and Dai, Bo},
booktitle = {Advances in Neural Information Processing Systems},
doi = {10.52202/085713-1482},
editor = {D. Belgrave and C. Zhang and H. Lin and R. Pascanu and P. Koniusz and M. Ghassemi and N. Chen},
pages = {44491--44544},
publisher = {Curran Associates, Inc.},
title = {Matryoshka Pilot: Learning to Drive Black-Box LLMs with LLMs},
url = {https://proceedings.neurips.cc/paper_files/paper/2025/file/3f055edce9cc3e90514b0716a16b37b4-Paper-Conference.pdf},
volume = {38, Main Conference},
year = {2025}
}