DigestAI news desk

Cut through the AI noise.

Agents & Tools1 min read

Google's Gemini 4 agents defeat Minecraft Ender Dragon without human input

Google researchers reported that four Gemini 4 Argon agents completed a full Minecraft survival run and killed the Ender Dragon entirely on their own. The agents each controlled a player via the open‑source Mineflayer library, communicated through in‑game chat, and dynamically split tasks such as tool making, food gathering, portal lighting and shelter building.

1 source

Key points

  • Four Gemini 4 Argon agents completed Minecraft survival and killed the Ender Dragon autonomously.
  • Agents exchanged 229 in‑game messages, wrote 2,783 programs, and died 605 times over 27.23 hours.
  • The run used 1.35 billion input tokens, costing $810 at list price ($1,620 at standard price).

The autonomous session lasted 27.23 hours. During that time the agents wrote and executed 2,783 programs, exchanged 229 messages, and died a total of 605 times before the final blow was delivered. The experiment consumed 1.35 billion input tokens, costing $810 at Gemini’s list price or $1,620 at the standard price. A fifth “Observer” agent monitored the run to ensure no cheating and to produce reports.

The team said this is the first end‑to‑end autonomous completion of Minecraft using only the standard Mineflayer library, without any pre‑scouted seeds or human‑written game‑specific tools, highlighting the growing capability of general‑purpose agentic AI models.

Model page: Gemini 4 Argon →

Full story from aol.com · by AOL · via Search: GoogleOpen source ↗

Google Gemini 4 Agents Just Beat Minecraft With No Human Intervention

aol.com · 19 October 2026

Loading the full article…

This text was published by aol.com and written by AOL. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗

Topics · follow one to build your own front page

The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.

Comments

via GitHub Discussions

More in Agents & Tools

All →

Related stories