Agentic LLM agents improve black-box optimization across domains

A Closer Look at Agentic BBO: Benchmarking LLM Agents for Black-Box Optimization

Machine LearningArtificial IntelligenceNeural and Evolutionary Computing

Summary

Black-box optimization means finding the best solution when you can only check results, and checking is expensive. The authors studied how large language model (LLM) agents can help solve these problems by combining language understanding with math tools. They created a benchmark that tests these agents on different tasks like tuning databases and designing molecules. Their tests showed that agentic LLMs generally do better than direct LLM methods and even beat some traditional numeric optimizers. They also found that the usefulness of math tools and task knowledge varies, and some of the gains come from combining LLM strategies with good numeric optimizers.

What this means in practice

  • For machine learning engineers: Optimize hyperparameters more effectively in limited evaluation budgets using agentic LLM-based optimization.
  • For chip design teams: Improve chip performance tuning by applying agentic LLMs that integrate design knowledge with optimization strategies.

Authors

Ming Chen, Rong-Xi Tan, Ke Xue, Yu-Jie Zhou, Taiye Lu, Zhi-Xuan Gao, Peng Xie, Zijun Shen, Chen Lu, Haopu Shang, Chao Qian

Abstract

Black-box optimization (BBO) arises in many scientific and engineering problems where objective evaluations are expensive and limited. Recent large language model (LLM) agents offer a new way to approach BBO by combining task semantics, computation, optimization tools, and feedback-driven decision making, showing great potential due to the integration with mathematically rigorous tools. However, existing agentic BBO studies use different task domains and system configurations, making their results difficult to compare and the effects of individual design choices hard to isolate. We therefore introduce AgenticBBO-Bench, a cross-domain benchmark for agentic BBO spanning synthetic functions, hyperparameter optimization, database tuning, chip design, and molecular design under a unified finite-budget evaluation protocol. In our experiments, agentic BBO achieves higher family-averaged scores than direct LLM-based methods in all five domains and outperforms the best numerical optimizers in four. We further study three factors shaping agent performance: optimization tools, task information and prior knowledge, and the role of the LLM during search. Our results show that additional numerical tools do not consistently improve performance, task semantics are broadly useful while more specific priors are less reliable, and numerical optimizers can effectively absorb gains from search trajectories established by the agent. Finally, we introduce a five-task frontier challenge within AgenticBBO-Bench and evaluate seven LLMs under the Codex agent harness, where GPT-6 Astra and DeepSeek-V4.1-Flash lie on the Pareto frontier of performance and cost among the evaluated models. Our code is available at https://github.com/lamda-bbo/agentic-bbo.