Show HN: Gemini 4 Argon plays StarCraft

StarSkirmish Bench v0.2 evaluates large language models by having them write StarCraft: Brood War bots in one hour, testing their long-horizon reasoning and agentic coding abilities. GPT-6 Astra achieved the highest score, followed by GPT-6.1 Sol, while Gemini 4 Argon, Claude Opus 5.5, and Claude Fable 5.1 scored similarly behind them, with all models still significantly outperformed by human-written bots. The analysis found that models practicing against the strongest bots during their hour correlated most strongly with higher scores.