Mine Odyssey: 180-Task Minecraft Benchmark Tests AI Agents' Real-World Spatial Navigation
junxian_he · x · 2026-10-09
Mine Odyssey is a benchmark for spatial agentic intelligence built on community-made Minecraft maps, covering 30 real-world locations (Notre Dame, Hong Kong, Versailles, the White House) across 180 tasks. Agents follow natural-language instructions to visit destinations in order, evaluated on success rate, checkpoint coverage, SPL, and turn counts. Community mods enable ablations on weather, HUDs, first/third-person views, and route lines, plus an Opus effort study.
More from coding & agent
- Claude Code Pushes Urgent Security Update Across All Installations — umeshonai · 2026-10-23
- Claude cut a 2.6-hour video export to 20 minutes — but only after being told to rebuild the server — byhewar · 2026-10-23
- 3 flight-booking rules my AI agent checks that I'd never do by hand — Efistoffeles · 2026-10-23
- OpenJet claims Qwen3.8-27B coding runs 1.9x faster than llama.cpp on one RTX 3090 — JeremyCMorgan · 2026-10-23
- Building an autonomous SEO agent in Hermes with three skills: programmatic pages, Keywords Everywhere, DataForSEO — EXM7777 · 2026-10-23
- Building a sports betting AI agent with onchain execution and human approval by default: architecture and lessons — Dry-Comfortable-2678 · 2026-10-23