Benchmark RadarGameEngineBench
GameEngineBench evaluates coding agents on scoped C++ implementation tasks within Unreal Engine 5 projects, built from nine real-world game repositories. The 110 tasks span gameplay, multiplayer, AI, animation, UI, and other areas, requiring native C++ changes that compile and pass behavioral tests.
- Released
- 2026-07-03
- Readiness
- Paper only
- Primary field
- Industrial & Engineering
Why it matters
Game-engine development presents unique challenges for coding agents, including stateful, real-time, and interactive systems. This benchmark fills a gap by focusing on deeply integrated C++ tasks, revealing limitations of current agents and providing a practical evaluation for real-world software engineering.
Motivation
Game engines provide real-time simulation, rendering, physics, interaction, networking, and asset pipelines, making them valuable not only for games but also for 3D applications in healthcare, robotics, architecture, manufacturing, and related domains.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.