Benchmark AI models by having them play StarCraft through code
StarSkirmish Bench has LLMs play StarCraft: Brood War by writing code, now refreshed with results from the flagship OpenAI, Anthropic, Google and Muse models including Gemini 4 Argon.
Source
__cayenne__ on Hacker News · 2026-10-16 · Project link
Written in our own words. Credit to the original author. Link checked 2026-10-16.
What they did
StarSkirmish Bench has LLMs play StarCraft: Brood War by writing code, now refreshed with results from the flagship OpenAI, Anthropic, Google and Muse models including Gemini 4 Argon.
Result (as claimed)
Public leaderboard; 143 points on Show HN.
Tools they used
Cost: Gemini 4 Argon: from $2 in / $10 out per 1M tokens.
Set it up
Gather your sources
Collect the papers, notes, links or files you want to work from in one place.
Load them into the tool
Open Gemini 4 Argon and add your sources. Keep the original files so you can check claims later.
Ask focused questions
Ask for summaries, comparisons or connections with citations to the source passages.
Verify and save
Spot check claims against the sources, then save the output as notes you can reuse.
Copy the full setup prompt
Paste into your agent and it will walk you through the setup, asking before anything that sends, posts or spends.
You are helping me set up this AI use case end to end in Claude Code. Goal: Benchmark AI models by having them play StarCraft through code What the original author did: StarSkirmish Bench has LLMs play StarCraft: Brood War by writing code, now refreshed with results from the flagship OpenAI, Anthropic, Google and Muse models including Gemini 4 Argon. Reported result (as claimed by the author): Public leaderboard; 143 points on Show HN. Source: https://news.ycombinator.com/item?id=50118688 Tools they used: Gemini 4 Argon Category: Research & learning Please: 1. Ask me 3 short questions about my situation (accounts I have, budget, what I want the output to look like). 2. Propose a step-by-step plan using these tools or cheaper alternatives, with the time and cost of each step. Use official pricing pages; say "check pricing" if unsure. 3. Then do the setup with me one step at a time, waiting for my OK before anything that sends messages, posts publicly, spends money or deletes data. 4. Finish with a checklist to run it again next time.
You are helping me set up this AI use case end to end in Codex. Goal: Benchmark AI models by having them play StarCraft through code What the original author did: StarSkirmish Bench has LLMs play StarCraft: Brood War by writing code, now refreshed with results from the flagship OpenAI, Anthropic, Google and Muse models including Gemini 4 Argon. Reported result (as claimed by the author): Public leaderboard; 143 points on Show HN. Source: https://news.ycombinator.com/item?id=50118688 Tools they used: Gemini 4 Argon Category: Research & learning Please: 1. Ask me 3 short questions about my situation (accounts I have, budget, what I want the output to look like). 2. Propose a step-by-step plan using these tools or cheaper alternatives, with the time and cost of each step. Use official pricing pages; say "check pricing" if unsure. 3. Then do the setup with me one step at a time, waiting for my OK before anything that sends messages, posts publicly, spends money or deletes data. 4. Finish with a checklist to run it again next time.
You are helping me set up this AI use case end to end in Grok Bot. Goal: Benchmark AI models by having them play StarCraft through code What the original author did: StarSkirmish Bench has LLMs play StarCraft: Brood War by writing code, now refreshed with results from the flagship OpenAI, Anthropic, Google and Muse models including Gemini 4 Argon. Reported result (as claimed by the author): Public leaderboard; 143 points on Show HN. Source: https://news.ycombinator.com/item?id=50118688 Tools they used: Gemini 4 Argon Category: Research & learning Please: 1. Ask me 3 short questions about my situation (accounts I have, budget, what I want the output to look like). 2. Propose a step-by-step plan using these tools or cheaper alternatives, with the time and cost of each step. Use official pricing pages; say "check pricing" if unsure. 3. Then do the setup with me one step at a time, waiting for my OK before anything that sends messages, posts publicly, spends money or deletes data. 4. Finish with a checklist to run it again next time.
The Daily 5
Get five new use cases tomorrow morning
Free. Real sources. Unsubscribe any time.
Related use cases
Benchmark AI decision models by making them play Pac-Man
Built Jevman so decision-model APIs compete at Pac-Man as a fast, simple decision benchmark.
Benchmark how much extra code AI models add
Built Parsimony, an open benchmark that scores how much unnecessary code models insert beyond solving the task.
Benchmark typed decision models with JevBench
Built JevBench, a reproducible benchmark for typed decision models used around agent memory and tooling.
Pull Reddit into Claude through your own MCP server
Built a personal MCP server so Claude can pull Reddit content directly instead of relying on brittle UI scraping.