Brood War Bench: Codex plays cheese, Opus wins battles, Grok too dumb to play
Kangwook_Lee · x · 2026-09-22
Last week benswerd launched Brood War Bench, pitting coding agents against each other in real-time StarCraft: Brood War matches, with results shared as a thread. Kangwook Lee amplified it, joking that he had "suddenly become an evaluation expert."
Key findings so far:
- Agents are still a long way from being able to annihilate the Terrans
- Codex relies on cheese strategies, confusing its opponents
- Grok is currently too dumb to play
- A shared clip shows Opus killing Fable after Fable over-invested in tech
Related event: Brood War Bench: Codex Tops with 100% Win Rate as Most LLMs Lose to Novices(3 posts)→
More from coding & agent
- Agent memory design question: what deserves a place in permanent memory? — Miserable-Shock3552 · 2026-09-22
- Dev demos near real-time editing with Gemini Flash Lite, plans open-source Chrome extension — Saboo_Shubham_ · 2026-09-22
- Local MLX vs hosted API: two AI agents battle in Chrome's T-Rex game — usamawahabkhan · 2026-09-22
- A silent watcher race dropped agent logs for a month: full post-mortem of a tailing bug — ClaudeCdGuy · 2026-09-22
- D-RAC cuts enterprise document chunking tokens by 95.7%, cost by up to 85.6% — Yellow-AI-NLP · 2026-09-22
- Did the Action Happen? Designing Confirmation Logic for Agent Tool Calls That Time Out — felix_baron · 2026-09-22