Frontier language models are deployed as black-box services, where model weights cannot be modified and customization is limited to prompting. We introduce Advisor Models, a method to train small open-weight models to generate dynamic, per-instance natural language advice that improves the capabilities of black-box frontier models. Advisor Models improve GPT-5.2's performance on RuleArena (Taxes) by 27.4%, reduce Gemini 3 Pro's steps taken in SWE agent tasks by 24.6%, and outperform static prompt optimizers in personalizing GPT-5 to user preferences (85-100% vs. 40-60%). We also find that advisors are transferable: an advisor trained with a low-cost student model still transfers improvements to a frontier model. Moreover, Advisor Models are robust: we observe no degradation on other benchmarks than the pipeline is trained on. Our method shows how to perform parametric optimization for black-box frontier models in a practical and cost-effective way.
@article{arxiv.2510.02453,
title = {How to Train Your Advisor: Steering Black-Box LLMs with Advisor Models},
author = {Parth Asawa and Alan Zhu and Abigail O'Neill and Matei Zaharia and Alexandros G. Dimakis and Joseph E. Gonzalez},
journal= {arXiv preprint arXiv:2510.02453},
year = {2026}
}
Comments
International Conference on Machine Learning (ICML) 2026