CAPO: Cost-Aware Prompt Optimization

Tom Zehle, Moritz Schlager, Timo Heiß, Matthias Feurer
Proceedings of the Fourth International Conference on Automated Machine Learning, PMLR 293:18/1-45, 2025.

Abstract

Large language models (LLMs) have revolutionized natural language processing by solving a wide range of tasks simply guided by a prompt. Yet their performance is highly sensitive to prompt formulation. While automatic prompt optimization addresses this challenge by finding optimal prompts, current methods require a substantial number of LLM calls and input tokens, making prompt optimization expensive. We introduce CAPO (Cost-Aware Prompt Optimization), an algorithm that enhances prompt optimization efficiency by integrating AutoML techniques. CAPO is an evolutionary approach with LLMs as operators, incorporating racing to save evaluations and multi-objective optimization to balance performance with prompt length. It jointly optimizes instructions and few-shot examples while leveraging task descriptions for improved robustness. Our extensive experiments across diverse datasets and LLMs demonstrate that CAPO outperforms state-of-the-art discrete prompt optimization methods in 11/15 cases with improvements up to 21% in accuracy. Our algorithm achieves better performances already with smaller budgets, saves evaluations through racing, and decreases average prompt length via a length penalty, making it both cost-efficient and cost-aware. Even without few-shot examples, CAPO outperforms its competitors and generally remains robust to initial prompts. CAPO represents an important step toward making prompt optimization more powerful and accessible by improving cost-efficiency.

Cite this Paper

BibTeX
@InProceedings{pmlr-v293-zehle25a, title = {CAPO: Cost-Aware Prompt Optimization}, author = {Zehle, Tom and Schlager, Moritz and Hei{\ss}, Timo and Feurer, Matthias}, booktitle = {Proceedings of the Fourth International Conference on Automated Machine Learning}, pages = {18/1--45}, year = {2025}, editor = {Akoglu, Leman and Doerr, Carola and van Rijn, Jan N. and Garnett, Roman and Gardner, Jacob R.}, volume = {293}, series = {Proceedings of Machine Learning Research}, month = {08--11 Sep}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v293/main/assets/zehle25a/zehle25a.pdf}, url = {https://proceedings.mlr.press/v293/zehle25a.html}, abstract = {Large language models (LLMs) have revolutionized natural language processing by solving a wide range of tasks simply guided by a prompt. Yet their performance is highly sensitive to prompt formulation. While automatic prompt optimization addresses this challenge by finding optimal prompts, current methods require a substantial number of LLM calls and input tokens, making prompt optimization expensive. We introduce CAPO (Cost-Aware Prompt Optimization), an algorithm that enhances prompt optimization efficiency by integrating AutoML techniques. CAPO is an evolutionary approach with LLMs as operators, incorporating racing to save evaluations and multi-objective optimization to balance performance with prompt length. It jointly optimizes instructions and few-shot examples while leveraging task descriptions for improved robustness. Our extensive experiments across diverse datasets and LLMs demonstrate that CAPO outperforms state-of-the-art discrete prompt optimization methods in 11/15 cases with improvements up to 21% in accuracy. Our algorithm achieves better performances already with smaller budgets, saves evaluations through racing, and decreases average prompt length via a length penalty, making it both cost-efficient and cost-aware. Even without few-shot examples, CAPO outperforms its competitors and generally remains robust to initial prompts. CAPO represents an important step toward making prompt optimization more powerful and accessible by improving cost-efficiency.} }
Endnote
%0 Conference Paper %T CAPO: Cost-Aware Prompt Optimization %A Tom Zehle %A Moritz Schlager %A Timo Heiß %A Matthias Feurer %B Proceedings of the Fourth International Conference on Automated Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Leman Akoglu %E Carola Doerr %E Jan N. van Rijn %E Roman Garnett %E Jacob R. Gardner %F pmlr-v293-zehle25a %I PMLR %P 18/1--45 %U https://proceedings.mlr.press/v293/zehle25a.html %V 293 %X Large language models (LLMs) have revolutionized natural language processing by solving a wide range of tasks simply guided by a prompt. Yet their performance is highly sensitive to prompt formulation. While automatic prompt optimization addresses this challenge by finding optimal prompts, current methods require a substantial number of LLM calls and input tokens, making prompt optimization expensive. We introduce CAPO (Cost-Aware Prompt Optimization), an algorithm that enhances prompt optimization efficiency by integrating AutoML techniques. CAPO is an evolutionary approach with LLMs as operators, incorporating racing to save evaluations and multi-objective optimization to balance performance with prompt length. It jointly optimizes instructions and few-shot examples while leveraging task descriptions for improved robustness. Our extensive experiments across diverse datasets and LLMs demonstrate that CAPO outperforms state-of-the-art discrete prompt optimization methods in 11/15 cases with improvements up to 21% in accuracy. Our algorithm achieves better performances already with smaller budgets, saves evaluations through racing, and decreases average prompt length via a length penalty, making it both cost-efficient and cost-aware. Even without few-shot examples, CAPO outperforms its competitors and generally remains robust to initial prompts. CAPO represents an important step toward making prompt optimization more powerful and accessible by improving cost-efficiency.
APA
Zehle, T., Schlager, M., Heiß, T. & Feurer, M.. (2025). CAPO: Cost-Aware Prompt Optimization. Proceedings of the Fourth International Conference on Automated Machine Learning, in Proceedings of Machine Learning Research 293:18/1-45 Available from https://proceedings.mlr.press/v293/zehle25a.html.

Related Material