GameASG-Bench: Benchmarking Autonomous Software Generation for Game Development
This paper provides a detailed benchmark for evaluating the performance of AI models in generating autonomous software for game development, which is relevant for anyone interested in the capabilities and limitations of these models.
We introduce GameASG-Bench, a benchmark for game development that evaluates autonomous software generation (ASG) models.
The benchmark includes 47 browser-native game-generation tasks across 12 genres and 2D/3D interaction.Experiments show that DeepSeek-V4-Flash achieves a 93.
2% L2 check pass rate but only a 55.3% strict task success rate.
The results highlight compliance gaps that high average check pass rates can hide.
Autonomous software generation (ASG) aims to turn human requirements into executable applications, but delivering these applications does not necessarily establish that their interacting components satisfy the specified behavioral requirements. We introduce GameASG-Bench, a benchmark that makes behavioral testability part of the generation task for game development. Our design declares an evaluation interface specification before generation, fixing legal starting scenarios, player-level actions, stable snapshots, rejection behavior, and invariants while leaving private implementations open. Concretely, we include: (i) static L1 checks that assess source-level compliance; and (ii) browser-executed L2 checks that combine semantic observations with real input and runtime evidence. We implement this protocol as 47 browser-native game-generation tasks spanning 12 primary genres and both 2D and 3D interaction, each with executable checks and an independently verified reference implementation. Our experiments answer four key questions about end-to-end agent performance, tool access and nominal turn budget, reasoning effort, and harness choice. Across nine agent stacks, the highest observed mean L2 check pass rate is 93.2%, yet the highest observed strict task success rate, requiring all L1 and applicable L2 prerequisite and core requirement checks, is only 55.3% (26/47 tasks). For DeepSeek-V4-Flash, full tool access and larger nominal turn budgets yield more strict task successes, while the strict task success rate is not monotonic in reasoning effort. Both tested harnesses achieve 18 strict task successes, but only ten tasks succeed under both. These results expose task-level compliance gaps that high average check pass rates actually obscure.
This report is categorized under AITOP's "Models" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows and US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include GameASG-Bench, autonomous software generation, game development, benchmark.
Primary source verification: published directly by arXiv: DeepSeek with first-party authority.
