EvoPolicyGym introduces benchmark for evaluating autonomous policy evolution

The authors introduce Autonomous Policy Evolution, a controlled evaluation setting where an agent repeatedly edits an executable policy system within a fixed interaction budget to iteratively improve explored policies.

• EvoPolicyGym is a benchmark built from compact interactive RL environments designed to assess this iterative improvement process.
• GPT-5.5 achieves the strongest aggregate rank score and top-two performance across all 16 environments in the suite.
• The benchmark provides trajectory-level diagnostics that distinguish how agents allocate budget and convert feedback into parametric tuning.

The authors argue that strong autonomous policy evolution depends on discovering task-appropriate mechanisms and refining policies under bounded feedback rather than just isolated task wins.