A StarCraft BW Arena where LLMs play by writing code

About A StarCraft BW Arena where LLMs play by writing code

StarSkirmish: where LLMs write code to play StarCraft.

In the maker’s words

StarSkirmish: Hillclimb is a long horizon implementation of the StarCraft: Broodwar based LLM benchmark that I released last week. Each LLM writes a Protoss bot in C++ against BWAPI, played on OpenBW. They must climb five tiers of Protoss opponents, from scripted demo bots up to top rated human written bots like Stardust and PurpleWave. The StarSkirmish bench restricted LLMs to 1 hour of wall clock time for reasoning. With Hillclimb there is no fixed time limit. GPT-6 Astra and Claude 5.5 Opus both benefit from substantially longer reasoning periods than 1 hour and so this format is a more realistic depiction of their abilities

Where people found it

• Hacker NewsShow HN: GPT-6 Astra and Claude 5.5 Opus race to create the best StarCraft bot2 points0 comments8 days ago
• Hacker NewsShow HN: A StarCraft BW Arena where LLMs play by writing code4 points1 comments14 days ago