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.
Source
felix089 on Hacker News · 2026-10-08 · Project link
Written in our own words. Credit to the original author. Link checked 2026-10-15.
What they did
Built Jevman so decision-model APIs compete at Pac-Man as a fast, simple decision benchmark.
Result (as claimed)
Public Pac-Man benchmark across several decision model endpoints.
Tools they used
The source does not name specific tools.
Cost: Pricing depends on the tools you pick; check each official pricing page.
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 the tools you choose 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 decision models by making them play Pac-Man What the original author did: Built Jevman so decision-model APIs compete at Pac-Man as a fast, simple decision benchmark. Reported result (as claimed by the author): Public Pac-Man benchmark across several decision model endpoints. Source: https://news.ycombinator.com/item?id=50007993 Tools they used: not named in the source 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 decision models by making them play Pac-Man What the original author did: Built Jevman so decision-model APIs compete at Pac-Man as a fast, simple decision benchmark. Reported result (as claimed by the author): Public Pac-Man benchmark across several decision model endpoints. Source: https://news.ycombinator.com/item?id=50007993 Tools they used: not named in the source 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 decision models by making them play Pac-Man What the original author did: Built Jevman so decision-model APIs compete at Pac-Man as a fast, simple decision benchmark. Reported result (as claimed by the author): Public Pac-Man benchmark across several decision model endpoints. Source: https://news.ycombinator.com/item?id=50007993 Tools they used: not named in the source 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 how much extra code AI models add
Built Parsimony, an open benchmark that scores how much unnecessary code models insert beyond solving the task.
Crawl the web for agents without an LLM in the crawl loop
Built Jevcrawl, an MCP and CLI crawler that finds answer pages more often without calling an LLM while crawling.
Prep for a presentation by listening to it
Upload a dense deck to NotebookLM and have it generate an AI podcast discussing it, then listen before you present.
Hunt for unknown planets in public telescope data
A user says he used Claude Code to analyze public data and found what appears to be a previously unknown planet.