RustEvo² introduces the first evolving benchmark to evaluate how well LLMs adapt to Rust's rapidly changing APIs across versions.

• Addresses critical gap in existing benchmarks by systematically evaluating LLM performance across API transitions
• Provides an automated framework to generate evolutionary test cases based on real Rust API changes
• Reveals significant challenges in version-aware code generation that current LLMs struggle to overcome

This research is crucial for engineering teams using AI code generation tools with evolving languages, highlighting the need for more adaptive models that can maintain compatibility with changing programming ecosystems.

RustEvo²: An Evolving Benchmark for API Evolution in LLM-based Rust Code Generation