Darwin Gödel Machine evolves its own coding agent
Sakana AI and University of British Columbia researchers released a coding agent that modifies its own Python code, evaluates descendants and preserves a diverse archive of prior variants. Reported SWE-bench performance rose from 20 percent to 50 percent during the research run.
Why it matters: Agent scaffolds can become objects of automated search, making evaluation design and containment as important as the underlying model.