New benchmark tests if AI agents can update software tests alongside code changes

Original: TestEvo-Bench: An Executable and Live Benchmark for Test and Code Co-Evolution

Source: arxiv.org ↗