# EvoPolicyGym: Evaluating Autonomous Policy Evolution in Interactive Environments

arXiv: https://arxiv.org/abs/2607.02440 · HTML: https://mlsanity.com/paper/2607.02440/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 21 of its 28 references that are in the map.
- First posted on arXiv 2 Jul 2026.

## Builds on (most cited first, 21 of 21)

- [Reflexion: Language Agents with Verbal Reinforcement Learning](https://mlsanity.com/paper/2303.11366/) (2023)
- [SWE-bench: Can Language Models Resolve Real-World GitHub Issues?](https://mlsanity.com/paper/2310.06770/) (2023)
- [Voyager: An Open-Ended Embodied Agent with Large Language Models](https://mlsanity.com/paper/2305.16291/) (2023)
- [WebArena: A Realistic Web Environment for Building Autonomous Agents](https://mlsanity.com/paper/2307.13854/) (2023)
- [SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineering](https://mlsanity.com/paper/2405.15793/) (2024)
- [AgentBench: Evaluating LLMs as Agents](https://mlsanity.com/paper/2308.03688/) (2023)
- [OSWorld: Benchmarking Multimodal Agents for Open-Ended Tasks in Real Computer Environments](https://mlsanity.com/paper/2404.07972/) (2024)
- [OpenHands: An Open Platform for AI Software Developers as Generalist Agents](https://mlsanity.com/paper/2407.16741/) (2024)
- [Eureka: Human-Level Reward Design via Coding Large Language Models](https://mlsanity.com/paper/2310.12931/) (2023)
- [MLE-bench: Evaluating Machine Learning Agents on Machine Learning Engineering](https://mlsanity.com/paper/2410.07095/) (2024)
- [MLAgentBench: Evaluating Language Agents on Machine Learning Experimentation](https://mlsanity.com/paper/2310.03302/) (2023)
- [WorkArena: How Capable Are Web Agents at Solving Common Knowledge Work Tasks?](https://mlsanity.com/paper/2403.07718/) (2024)
- [Minigrid & Miniworld: Modular & Customizable Reinforcement Learning Environments for Goal-Oriented Tasks](https://mlsanity.com/paper/2306.13831/) (2023)
- [SWE-EVO: Benchmarking Coding Agents in Long-Horizon Software Evolution Scenarios](https://mlsanity.com/paper/2512.18470/) (2025)
- [SWE-chat: Coding Agent Interactions From Real Users in the Wild](https://mlsanity.com/paper/2604.20779/) (2026)
- [Understanding Code Agent Behaviour: An Empirical Study of Success and Failure Trajectories](https://mlsanity.com/paper/2511.00197/) (2025)
- [SlopCodeBench: Benchmarking How Coding Agents Degrade Over Long-Horizon Iterative Tasks](https://mlsanity.com/paper/2603.24755/) (2026)
- [AgentLens: Visual Analysis for Agent Behaviors in LLM-based Autonomous Systems](https://mlsanity.com/paper/2402.08995/) (2024)
- [How Coding Agents Fail Their Users: A Large-Scale Analysis of Developer-Agent Misalignment in 20,574 Real-World Sessions](https://mlsanity.com/paper/2605.29442/) (2026)
- [SpecBench: Evaluating Specification-Level Reasoning for Software Engineering LLM Agents](https://mlsanity.com/paper/2605.30314/) (2026)
- [RoadmapBench: Evaluating Long-Horizon Agentic Software Development Across Version Upgrades](https://mlsanity.com/paper/2605.15846/) (2026)

## Built on by (most cited first, 2 of 2)

- [When Better Gets Worse: Improvement Fidelity for Self-Improving Agents in Adaptive Worlds](https://mlsanity.com/paper/2609.32677/) (2026)
- [Can AI Agents Learn Their Way to the Top? Evaluating Heuristic Learning in a Long-Running Game Agent Competition](https://mlsanity.com/paper/2610.12341/) (2026)

## Abstract (arXiv)

Autonomous agents are increasingly expected to improve executable policies through feedback, yet existing evaluations often collapse this process into a final score or confound it with open-ended software-engineering progress. We introduce Autonomous Policy Evolution, a controlled evaluation setting in which a harness-model agent repeatedly edits an executable policy system under a fixed interaction budget. We instantiate this setting in EvoPolicyGym, a benchmark built from compact interactive RL environments that evaluates how agents iteratively improve explored policies. On the EvoPolicyGym suite, GPT-5.5 achieves the strongest aggregate rank score and top-two performance on all 16 environments. Beyond leaderboard results, EvoPolicyGym also provides trajectory-level diagnostics that distinguish how agents allocate budget, convert feedback into parametric tuning. These analyses show that strong autonomous policy evolution depends not only on isolated task wins, but on discovering task-appropriate mechanisms and refining policies under bounded feedback.