Google AI agents beat Minecraft's Ender Dragon without human help
Four Gemini 4 Argon agents from Google DeepMind finished Minecraft after about 27 hours of autonomous play. The reports describe the feat as a world first.
By Standard Topic. AI-generated from the reporting listed under Coverage. Published . Updated .
This story is no longer being updated. It reflects reporting up to .
What happened
Four autonomous AI agents powered by Google DeepMind's Gemini 4 Argon model completed Minecraft from scratch, defeating the Ender Dragon. The effort took just over 27 hours, and the agents operated without human assistance. Google researchers released a paper announcing the result.
Why it matters
The reports describe the result as another major gaming benchmark for Google DeepMind. They also say it is believed to be a world first.
Timeline
- Oct 19, 6:09 AM ET: Four Gemini 4 AI agents completed Minecraft and defeated the Ender Dragon after operating for more than a day without human assistance.
- Oct 19, 6:56 AM ET: Google researchers released a paper claiming the four agents beat Minecraft, taking just over 27 hours to defeat the Ender Dragon.
- Oct 20, 7:47 AM ET: Reports said the agents defeated the Ender Dragon after 27 hours and 14 minutes, with 605 deaths recorded along the way.
Coverage
- Google DeepMind AI Agents Beat Minecraft... After 27 Hours And 605 'Deaths' (theaiinsider.tech)
- Google DeepMind’s AI wins another game, company says Gemini 4 Argon agents beat Minecraft from scratch in 27 hours (Times of India)
- It Took 4 Google AI Agents Over 27 Hours to Defeat the Ender Dragon in Minecraft Without Help From Humans (Kirk McKeand)
- 4 Google AI agents defeat Minecraft boss without human assistance - Root-Nation.com (root-nation.com)
- Four Gemini 4 Argon agents beat Minecraft without any human help (the-decoder.com)
More from Technology & AI
- Eight states sue OpenAI, alleging misleading AI safety claims
- OpenAI announces solution to Navier-Stokes math problem
- Datadog bets AI agents can handle testing of AI-written code
- Anthropic and OpenAI sketch wealth-sharing plans before IPOs
- Mamdani administration proposes AI customer service rules
- Florida engineers study how AI behaves in high-risk situations