@inproceedings{kim-etal-2026-beyond,
title = "Beyond Perfect {API}s: Evaluating {LLM} Agents Under Realistic {API} Complexity",
author = "Kim, Doyoung and
Ren, Zhiwei and
Hao, Jie and
Sun, Zhongkai and
Wang, Lichao and
Ma, Xiyao and
Han, Xu and
Yin, Jun and
Ji, Heng and
Shen, Wei and
Fan, Xing and
Yao, Benjamin Z. and
Guo, Chenlei",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.1965/",
pages = "44744--44774",
ISBN = "979-8-89176-459-0",
abstract = "The rise of fully autonomous agents necessitates a rigorous evaluation of their reliability under **real-world API complexity**. Accordingly, moving beyond perfectly **sanitized** settings that disregard realistic obstacles, we introduce ***WildAGTEval***, a configurable benchmark environment that accounts for two critical dimensions of realistic API complexity: **API specification**, which includes detailed documentation and usage constraints, and **API execution**, which captures runtime challenges. Consequently, ***WildAGTEval*** offers (i) an API system spanning 86 functions across 7 domains with 60 independently activatable complexity scenarios, yielding approximately 32K distinct test configurations, and (ii) 300 multi-turn conversations grounded in this system. Through systematic evaluation of 11 state-of-the-art LLMs, we find that: (1) individual complexities reduce accuracy by up to 23.1{\%}; (2) compounding them amplifies this drop to 52.1{\%}; and crucially, (3) failing agents fabricate success and distort intent rather than admitting limitations, exposing a critical reliability gap invisible to existing benchmarks."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="kim-etal-2026-beyond">
<titleInfo>
<title>Beyond Perfect APIs: Evaluating LLM Agents Under Realistic API Complexity</title>
</titleInfo>
<name type="personal">
<namePart type="given">Doyoung</namePart>
<namePart type="family">Kim</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhiwei</namePart>
<namePart type="family">Ren</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jie</namePart>
<namePart type="family">Hao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhongkai</namePart>
<namePart type="family">Sun</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lichao</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiyao</namePart>
<namePart type="family">Ma</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xu</namePart>
<namePart type="family">Han</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jun</namePart>
<namePart type="family">Yin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Heng</namePart>
<namePart type="family">Ji</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wei</namePart>
<namePart type="family">Shen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xing</namePart>
<namePart type="family">Fan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Benjamin</namePart>
<namePart type="given">Z</namePart>
<namePart type="family">Yao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chenlei</namePart>
<namePart type="family">Guo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>The rise of fully autonomous agents necessitates a rigorous evaluation of their reliability under **real-world API complexity**. Accordingly, moving beyond perfectly **sanitized** settings that disregard realistic obstacles, we introduce ***WildAGTEval***, a configurable benchmark environment that accounts for two critical dimensions of realistic API complexity: **API specification**, which includes detailed documentation and usage constraints, and **API execution**, which captures runtime challenges. Consequently, ***WildAGTEval*** offers (i) an API system spanning 86 functions across 7 domains with 60 independently activatable complexity scenarios, yielding approximately 32K distinct test configurations, and (ii) 300 multi-turn conversations grounded in this system. Through systematic evaluation of 11 state-of-the-art LLMs, we find that: (1) individual complexities reduce accuracy by up to 23.1%; (2) compounding them amplifies this drop to 52.1%; and crucially, (3) failing agents fabricate success and distort intent rather than admitting limitations, exposing a critical reliability gap invisible to existing benchmarks.</abstract>
<identifier type="citekey">kim-etal-2026-beyond</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.1965/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>44744</start>
<end>44774</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Beyond Perfect APIs: Evaluating LLM Agents Under Realistic API Complexity
%A Kim, Doyoung
%A Ren, Zhiwei
%A Hao, Jie
%A Sun, Zhongkai
%A Wang, Lichao
%A Ma, Xiyao
%A Han, Xu
%A Yin, Jun
%A Ji, Heng
%A Shen, Wei
%A Fan, Xing
%A Yao, Benjamin Z.
%A Guo, Chenlei
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F kim-etal-2026-beyond
%X The rise of fully autonomous agents necessitates a rigorous evaluation of their reliability under **real-world API complexity**. Accordingly, moving beyond perfectly **sanitized** settings that disregard realistic obstacles, we introduce ***WildAGTEval***, a configurable benchmark environment that accounts for two critical dimensions of realistic API complexity: **API specification**, which includes detailed documentation and usage constraints, and **API execution**, which captures runtime challenges. Consequently, ***WildAGTEval*** offers (i) an API system spanning 86 functions across 7 domains with 60 independently activatable complexity scenarios, yielding approximately 32K distinct test configurations, and (ii) 300 multi-turn conversations grounded in this system. Through systematic evaluation of 11 state-of-the-art LLMs, we find that: (1) individual complexities reduce accuracy by up to 23.1%; (2) compounding them amplifies this drop to 52.1%; and crucially, (3) failing agents fabricate success and distort intent rather than admitting limitations, exposing a critical reliability gap invisible to existing benchmarks.
%U https://aclanthology.org/2026.emnlp-main.1965/
%P 44744-44774
Markdown (Informal)
[Beyond Perfect APIs: Evaluating LLM Agents Under Realistic API Complexity](https://aclanthology.org/2026.emnlp-main.1965/) (Kim et al., EMNLP 2026)
ACL
- Doyoung Kim, Zhiwei Ren, Jie Hao, Zhongkai Sun, Lichao Wang, Xiyao Ma, Xu Han, Jun Yin, Heng Ji, Wei Shen, Xing Fan, Benjamin Z. Yao, and Chenlei Guo. 2026. Beyond Perfect APIs: Evaluating LLM Agents Under Realistic API Complexity. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 44744–44774, Budapest, Hungary. Association for Computational Linguistics.