SysMoBench: Evaluating AI on Formally Modeling Complex Real-World Systems

2025-09-27

AI Summary

This work addresses the lack of rigorous evaluation of large language models’ (LLMs) capability to formally model complex real-world systems—particularly concurrent and distributed systems. We introduce the first benchmark for assessing AI-driven formal modeling competence on large-scale, production-grade systems. Methodologically, we adopt TLA+ as the specification language and integrate static analysis, model checking, and runtime verification to enable automated, multi-dimensional assessment across four criteria: syntactic correctness, behavioral consistency, code-to-specification fidelity, and invariant validity. The benchmark encompasses nine representative system components—including Etcd, Redis, and Asterinas OS—and covers core concurrency abstractions such as the Raft consensus protocol and OS synchronization primitives. Experimental results expose systematic deficiencies in current LLMs’ ability to abstract deep concurrency semantics and rigorously preserve invariants. Our benchmark provides a reproducible empirical foundation and an extensible evaluation framework for AI-augmented formal methods research.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Cognitive Modeling & Cognitive Systems: Agent Architectures

Application Category

• 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
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

Formal models are essential to specifying large, complex computer systems and verifying their correctness, but are notoriously expensive to write and maintain. Recent advances in generative AI show promise in generating certain forms of specifications. However, existing work mostly targets small code, not complete systems. It is unclear whether AI can deal with realistic system artifacts, as this requires abstracting their complex behavioral properties into formal models. We present SysMoBench, a benchmark that evaluates AI's ability to formally model large, complex systems. We focus on concurrent and distributed systems, which are keystones of today's critical computing infrastructures, encompassing operating systems and cloud infrastructure. We use TLA+, the it de facto specification language for concurrent and distributed systems, though the benchmark can be extended to other specification languages. We address the primary challenge of evaluating AI-generated models by automating metrics like syntactic and runtime correctness, conformance to system code, and invariant correctness. SysMoBench currently includes nine diverse system artifacts: the Raft implementation of Etcd and Redis, the Spinlock and Mutex in Asterinas OS, etc.; more artifacts are being actively added. SysMoBench enables us to understand the capabilities and limitations of today's LLMs and agents, putting tools in this area on a firm footing and opening up promising new research directions.

Problem

• Evaluating AI's ability to formally model large complex systems
• Assessing AI-generated formal specifications for concurrent distributed systems
• Automating correctness metrics for AI-generated system behavior models

Innovation

• Automated metrics for evaluating AI-generated formal models
• Benchmark for modeling concurrent and distributed systems
• Using TLA+ specification language for system verification