Saved in:
Bibliographic Details
Main Authors: Chi, Yizhe, Hong, Deyao, Jiang, Dapeng, Luo, Tianwei, Yang, Kaisen, Zhang, Boshi, Cao, Zhe, Fan, Xiaoyan, He, Bingxiang, Hao, Han, Jin, Weiyang, Lei, Dianqiao, Liu, Qingle, Qian, Houde, Wang, Bowen, Wang, Situ, Zheng, Youjie, Zhou, Yifan, Xiao, Calvin, Cai, Eren, Na, Qinhuai
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.12290
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908994893774848
author Chi, Yizhe
Hong, Deyao
Jiang, Dapeng
Luo, Tianwei
Yang, Kaisen
Zhang, Boshi
Cao, Zhe
Fan, Xiaoyan
He, Bingxiang
Hao, Han
Jin, Weiyang
Lei, Dianqiao
Liu, Qingle
Qian, Houde
Wang, Bowen
Wang, Situ
Zheng, Youjie
Zhou, Yifan
Xiao, Calvin
Cai, Eren
Na, Qinhuai
author_facet Chi, Yizhe
Hong, Deyao
Jiang, Dapeng
Luo, Tianwei
Yang, Kaisen
Zhang, Boshi
Cao, Zhe
Fan, Xiaoyan
He, Bingxiang
Hao, Han
Jin, Weiyang
Lei, Dianqiao
Liu, Qingle
Qian, Houde
Wang, Bowen
Wang, Situ
Zheng, Youjie
Zhou, Yifan
Xiao, Calvin
Cai, Eren
Na, Qinhuai
contents Current LLM agent benchmarks, which predominantly focus on binary pass/fail tasks such as code generation or search-based question answering, often neglect the value of real-world engineering that is often captured through the iterative optimization of feasible designs. To this end, we introduce Frontier-Eng, a human-verified benchmark for generative optimization -- an iterative propose-execute-evaluate loop in which an agent generates candidate artifacts, receives executable verifier feedback, and revises them under a fixed interaction budget -- spanning $47$ tasks across five broad engineering categories. Unlike previous suites, Frontier-Eng tasks are grounded in industrial-grade simulators and verifiers that provide continuous reward signals and enforce hard feasibility constraints under constrained budgets. We evaluate eight frontier language models using representative search frameworks, finding that while GPT 5.4 achieves the most robust performance, the benchmark remains challenging for all models. Our analysis suggests a dual power-law decay in improvement frequency ($\sim$ 1/iteration) and magnitude ($\sim$ 1/improvement count). We further show that although width improves parallelism and diversity, depth remains crucial for hard-won improvements under a fixed budget. Frontier-Eng establishes a new standard for assessing the capacity of AI agents to integrate domain knowledge with executable feedback to solve complex, open-ended engineering problems.
format Preprint
id arxiv_https___arxiv_org_abs_2604_12290
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Frontier-Eng: Benchmarking Self-Evolving Agents on Real-World Engineering Tasks with Generative Optimization
Chi, Yizhe
Hong, Deyao
Jiang, Dapeng
Luo, Tianwei
Yang, Kaisen
Zhang, Boshi
Cao, Zhe
Fan, Xiaoyan
He, Bingxiang
Hao, Han
Jin, Weiyang
Lei, Dianqiao
Liu, Qingle
Qian, Houde
Wang, Bowen
Wang, Situ
Zheng, Youjie
Zhou, Yifan
Xiao, Calvin
Cai, Eren
Na, Qinhuai
Artificial Intelligence
Computation and Language
Current LLM agent benchmarks, which predominantly focus on binary pass/fail tasks such as code generation or search-based question answering, often neglect the value of real-world engineering that is often captured through the iterative optimization of feasible designs. To this end, we introduce Frontier-Eng, a human-verified benchmark for generative optimization -- an iterative propose-execute-evaluate loop in which an agent generates candidate artifacts, receives executable verifier feedback, and revises them under a fixed interaction budget -- spanning $47$ tasks across five broad engineering categories. Unlike previous suites, Frontier-Eng tasks are grounded in industrial-grade simulators and verifiers that provide continuous reward signals and enforce hard feasibility constraints under constrained budgets. We evaluate eight frontier language models using representative search frameworks, finding that while GPT 5.4 achieves the most robust performance, the benchmark remains challenging for all models. Our analysis suggests a dual power-law decay in improvement frequency ($\sim$ 1/iteration) and magnitude ($\sim$ 1/improvement count). We further show that although width improves parallelism and diversity, depth remains crucial for hard-won improvements under a fixed budget. Frontier-Eng establishes a new standard for assessing the capacity of AI agents to integrate domain knowledge with executable feedback to solve complex, open-ended engineering problems.
title Frontier-Eng: Benchmarking Self-Evolving Agents on Real-World Engineering Tasks with Generative Optimization
topic Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2604.12290