Researchers have introduced 3D GameDev Arena, a new benchmark designed to evaluate the capabilities of coding agents in developing complex 3D browser games. This benchmark comprises 108 distinct game development tasks across 14 genres, requiring agents to plan, implement, and test games from scratch based on text specifications and reference images. The evaluation process includes objective code-level checks, agent-based quality assurance for playability and presentation, and human preference testing in a public arena. Initial findings suggest that current coding agents still struggle with long-horizon 3D game development, and code correctness alone does not guarantee player satisfaction. AI
IMPACT This benchmark will drive progress in AI agents' ability to handle complex, long-horizon tasks like game development.
RANK_REASON The cluster describes a new research paper introducing a benchmark for AI agents. [lever_c_demoted from research: ic=1 ai=1.0]
- 3D GameDev Arena
- alphaXiv
- arXiv
- Bradley--Terry model
- CatalyzeX
- DagsHub
- Gotit.pub
- Hugging Face
- Influence Flower
- ScienceCast
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →