Google DeepMind says four Gemini 4 Argon agents beat Minecraft without any human help in 27 hours and 14 minutes, costing $1,620 at Argon's planned 2027 price
The Google Deepmind team dropped the agents into a randomly generated world without assigning roles, strategies, or game-specific tools.
The experiment arrives days after Argon reached Gemini’s paid API tier, turning a research demonstration into a concrete measure of what a multi-agent workload may cost at its planned 2027 price. DeepMind says the agents coordinated through chat, wrote code, shared notes and debugged without assigned roles or game-specific tools.
First-order effects
Google DeepMind gains an end-to-end demonstration for Gemini 4 Argon: four agents can divide work and complete a long-horizon task in a randomized 3D environment at a stated estimated run cost of $1,620.
Developers evaluating Argon receive a visible benchmark for code-driven, multi-agent orchestration rather than a single-model prompt-response task.
Second-order effects
Competing frontier-model providers face a more demanding comparison point: performance must include coordination, persistence and the cost of repeated agent actions, not only a model’s isolated output.
Agent builders can use the reported cost and runtime as a reference for deciding which long-running tasks can justify multi-agent inference budgets.
Third-order effects
If comparable results transfer beyond game environments, agent evaluation will shift toward teams that plan, communicate and recover from errors over long task horizons rather than single-agent benchmarks.
The economic bottleneck for agentic systems will increasingly be the total inference cost of coordination and retries, making model pricing and runtime efficiency product-level differentiators.
The trend: Frontier-model competition is moving from standalone answers toward economically measurable teams of agents that coordinate over extended computer-mediated tasks.
Four Gemini 4 Argon agents beat Minecraft from scratch in 27 hours. Operating in a sandbox using only code and text, they wrote 2,783 programs, divided up jobs over chat, shared notes, and debugged errors together - all with no human help ↓ https://x.com/...
Four Gemini 4 Argon agents teamed up in Minecraft to defeat the Ender Dragon in 27 hours, writing every move as code and coordinating in the game's chat. Great work led by @maxencefaldor with Satinder Singh. Read more in our technical report: https://arxiv.org/....
The paper on how a team of Gemini 4 agents beat Minecraft is worth a look. It is accessible example (partially because it is AI written) of spontaneous coordination among agents to accomplish a complicated goal & how their approach differs from humans/ https://arxiv.org/... [imag…
An independent research experiment, not an official Minecraft milestone, and not approved by or associated with Mojang or Microsoft.Visuals use the Pixel Perfection pack by XSSheep & Nova_Wostra (CC BY-SA 4.0).
A groundbreaking Frontier result! @maxencefaldor and @robertarail did a phenomenal job in showing the Argon capabilities so effectively. What a model and what a team in every sense of the word!
Gemini 4 Argon agents in Minecraft collaborating, demonstrating remarkable long-term planning and problem-solving all the way to the end! Fascinating breakthrough. Congratulations to @maxencefaldor, @robertarail and Satinder Singh!
Congratulations to @maxencefaldor, @robertarail, Satinder Singh and the rest of the team! Four Gemini 4 Argon agents teamed up in Minecraft, writing code and sharing learnings for 27 hours to beat the Ender Dragon. A fantastic example of Argon's ability to plan, adapt, and solve …