# OmniGameArena: A Unified UE5 Benchmark for VLM Game Agents with Improvement Dynamics

arXiv: https://arxiv.org/abs/2606.09826 · HTML: https://mlsanity.com/paper/2606.09826/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Placed in Reinforcement learning for LLM agents, one of 16 directions in LLM agents and tool use.
- Builds on 17 of its 28 references that are in the map.
- First posted on arXiv 8 Jun 2026.

## Placed in

- LLM agents and tool use / [Reinforcement learning for LLM agents](https://mlsanity.com/agents/reinforcement-learning-for-llm-agents/)

## Builds on (most cited first, 17 of 17)

- [ExpeL: LLM Agents Are Experiential Learners](https://mlsanity.com/paper/2308.10144/) (2023)
- [The NetHack Learning Environment](https://mlsanity.com/paper/2006.13760/) (2020)
- [Interactive Fiction Games: A Colossal Adventure](https://mlsanity.com/paper/1909.05398/) (2019)
- [SmartPlay: A Benchmark for LLMs as Intelligent Agents](https://mlsanity.com/paper/2310.01557/) (2023)
- [ChessGPT: Bridging Policy Learning and Language Modeling](https://mlsanity.com/paper/2306.09200/) (2023)
- [lmgame-Bench: How Good are LLMs at Playing Games?](https://mlsanity.com/paper/2505.15146/) (2025)
- [VideoGameBench: Can Vision-Language Models complete popular video games?](https://mlsanity.com/paper/2505.18134/) (2025)
- [GameArena: Evaluating LLM Reasoning through Live Computer Games](https://mlsanity.com/paper/2412.06394/) (2024)
- [NitroGen: An Open Foundation Model for Generalist Gaming Agents](https://mlsanity.com/paper/2601.02427/) (2026)
- [Game-TARS: Pretrained Foundation Models for Scalable Generalist Multimodal Game Agents](https://mlsanity.com/paper/2510.23691/) (2025)
- [V-ReasonBench: Toward Unified Reasoning Benchmark Suite for Video Generation Models](https://mlsanity.com/paper/2511.16668/) (2025)
- [Are Large Vision Language Models Good Game Players?](https://mlsanity.com/paper/2503.02358/) (2025)
- [WebGym: Scaling Training Environments for Visual Web Agents with Realistic Tasks](https://mlsanity.com/paper/2601.02439/) (2026)
- [Scaling Behavior Cloning Improves Causal Reasoning: An Open Model for Real-Time Video Game Playing](https://mlsanity.com/paper/2601.04575/) (2026)
- [GameVerse: Can Vision-Language Models Learn from Video-based Reflection?](https://mlsanity.com/paper/2603.06656/) (2026)
- [EmbRACE-3K: Embodied Reasoning and Action in Complex Environments](https://mlsanity.com/paper/2507.10548/) (2025)
- [AssetFormer: Modular 3D Assets Generation with Autoregressive Transformer](https://mlsanity.com/paper/2602.12100/) (2026)

## Built on by (most cited first, 4 of 4)

- [AI for Games in the Foundation Model Era](https://mlsanity.com/paper/2609.16679/) (2026)
- [DeliveryGym: An RL Environment for Long-Horizon Embodied Agent Planning with Adaptive Curriculum](https://mlsanity.com/paper/2609.19801/) (2026)
- [RTSGameBench: An RTS Benchmark for Strategic Reasoning by Vision-Language Models](https://mlsanity.com/paper/2606.18950/) (2026)
- [A2Z GameSpec-Bench: How Faithfully Can Coding Agents Generate Games from Game Design Specifications?](https://mlsanity.com/paper/2609.39564/) (2026)

## Abstract (arXiv)

Vision-language model (VLM) agents are increasingly deployed in interactive game environments. Yet game benchmarks for VLM agents typically report a single first-attempt score per (agent, game) pair, focus on single-agent Solo play, and lack unified protocols for evaluating heterogeneous agent classes (commercial VLMs, open-weight VLMs, and specialized game policies) on the same footing. We address these gaps with OmniGameArena, a real-time benchmark of twelve newly built Unreal Engine 5 games spanning Solo (7), PvP (3), and Coop (2) with unified action interfaces, and the Improvement Dynamics Curve (IDC), an agentic-reflection harness in which a tool-using reflector LLM autonomously refines a bounded skill prompt across multiple rounds. Beyond cold-start leaderboard scores, IDC exposes two additional observables for each (agent, game) pair: how the score evolves across reflection rounds, and how the learned skill behaves on held-out task variants. We report these observables for twelve VLM agents on the cold-start leaderboard and four top agents under IDC.