Google DeepMind paper: 4 general-purpose coding agents beat Minecraft without human intervention
A new Google DeepMind paper shows that 4 general-purpose coding agents beat Minecraft without human intervention. Each Gemini 4 Argon agent received only a goal (defeat the Ender Dragon) and a standard library for converting JavaScript into game actions; no one assigned them roles or wrote game tools for them, and the team organized itself through chat and shared notes. On Easy difficulty in a random world, they killed the Ender Dragon in 27 hours and 14 minutes, writing a total of 2,783 programs and dying 605 times; because this was only a single run, the success rate of this approach remains unknown.
Why read this
AI score 74
The paper shows general-purpose coding agents dividing up work among themselves to accomplish a goal in a long-horizon task, and provides the specific elapsed time and number of programs from a single run.
Topics
Body
New Google DeepMind paper shows 4 general-purpose coding agents beating Minecraft with no human help, writing every program and splitting the work themselves.
Finishing Minecraft means holding 1 plan across thousands of dependent steps, the same skill long software and business tasks need. Earlier Minecraft agents were trained on gameplay or handed game-specific planners.
Each Gemini 4 Argon agent got 1 goal, defeat the Ender Dragon, and a standard library that turns JavaScript into game actions. Nobody assigned roles or wrote game tools, so the team organized itself through chat and shared notes.
From a random world on Easy difficulty, they killed the dragon in 27 hours and 14 minutes, writing 2,783 programs and dying 605 times. It was a single run, so how often this setup succeeds is still unknown.
Before writing custom planners for a long task, try general coding agents with a plain API and a shared memory.