Gemini 4 Argon plays StarCraft
I've refreshed StarSkirmish Bench, a benchmark where LLMs play StarCraft BW by writing code, with results from the flagship models of OpenAI, Anthropic, Google, and Muse.
Reactions (1)
Similar products
by description similarity
All games are playable in the browser, and were written in the past month in AGC/Macro1 assembler, with the source code available on Github: https://github.com/1888games/2048_Apollo https://github.com/1888games/2DMaze_AGC https://github.com/1888games/1nvader-PDP-1
It's a unique way to watch StarCraft replays.
LocalScore
localscore.aiLocal LLM Benchmark
Speed Miners
speedminers.funA tiny RTS resource mini-game
ML, Biotech, Hardware, and Coordination Problems. Sometimes I write about hard problems and how to solve them.