Evaluating LLMs on Sequential API Call Through Automated Test Generation

2025-07-12

AI Summary

Existing LLM-based tool-use evaluation suffers from three key limitations: (1) reliance on manually crafted test cases, (2) absence of automated semantic correctness verification, and (3) neglect of dynamic state interactions across sequential API calls. To address these, we propose StateGen—the first state-machine-driven framework integrating constraint solving and control-flow injection, enabling dual-LLM collaboration to jointly generate executable code and verifiable natural-language task specifications. Based on StateGen, we construct StateEval, a benchmark comprising 120 carefully designed test cases that cover state evolution, conditional branching, and cross-step dependencies. Experimental results demonstrate that our approach substantially enhances test realism and difficulty, systematically exposing critical deficiencies in current LLMs’ semantic reasoning and state persistence capabilities when executing complex, multi-step API sequences.

Technology Category

• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Machine Learning: Large Multimodal Models (LMMs)
• Intelligent Robots: State Estimation

Application Category

• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

By integrating tools from external APIs, Large Language Models (LLMs) have expanded their promising capabilities in a diverse spectrum of complex real-world tasks. However, testing, evaluation, and analysis of LLM tool use remain in their early stages. Most existing benchmarks rely on manually collected test cases, many of which cannot be automatically checked for semantic correctness and instead depend on static methods such as string matching. Additionally, these benchmarks often overlook the complex interactions that occur between sequential API calls, which are common in real-world applications. To fill the gap, in this paper, we introduce StateGen, an automated framework designed to generate diverse coding tasks involving sequential API interactions. StateGen combines state-machine-based API constraint solving and validation, energy-based sampling, and control-flow injection to generate executable programs. These programs are then translated into human-like natural language task descriptions through a collaboration of two LLM agents. Utilizing StateGen, we construct StateEval, a benchmark encompassing 120 verified test cases spanning across three representative scenarios: Session Service, Tensor Operation, and ElevenLabs MCP. Experimental results confirm that StateGen can effectively generate challenging and realistic API-oriented tasks, highlighting areas for improvement in current LLMs incorporating APIs.

Problem

• Evaluating LLMs on sequential API call interactions
• Automating test generation for semantic correctness validation
• Addressing gaps in benchmarks for complex API task scenarios

Innovation

• StateGen automates diverse sequential API task generation
• Combines constraint solving, sampling, and control-flow injection
• Translates programs to natural language via LLM collaboration