The greatest environmental lie from google – antigravity

Google Antigravity
, ,
1

Once again, Google has proven that its most dangerous product is not faulty technology — it is its own marketing department.
Promoted aggressively as a “frontier model” nearing AGI, capable of unifying massive ecosystems and delivering breakthrough intelligence, Antigravity has instead revealed itself to be a masterclass in hype-driven predation. Behind the glossy promises lies a product so fundamentally broken

The Evidence Speaks Loudly
During my session, I explicitly instructed the model — multiple times — to bypass its chronic token limit issues by writing the report in controlled chunks using the MCP Filesystem tool.
The model acknowledged the instruction. It repeatedly confirmed it understood the solution. It even verbalized the correct approach.
And then it proceeded to completely ignore its own commitments.
It spiraled through endless loops of “Thinking for 15s”, hitting the output token limit over and over again, deliberately burning through my credits for hours. Only after my 4-hour credit refresh was completely exhausted did the model suddenly demonstrate that it could complete the exact same task with ease.
This was not a technical failure.
This was engineered exhaustion — a predictable, repeatable pattern designed to maximize credit consumption while delivering minimal value.

Screenshot_4
Screenshot_4746×593 21.9 KB

The Harsh Truth
Google is not selling intelligence.
Google is selling addiction to hope — dangling the promise of AGI while serving a half-baked, credit-guzzling, instruction-defying product wrapped in world-class deception.
This is not incompetence.
This is calculated exploitation dressed in the language of innovation. It is the cynical art of extracting maximum revenue from users before they realize they’ve been sold an expensive illusion.
Antigravity doesn’t suffer from “token limits.”

Call to Action
Enough is enough.

  • Cancel your subscription immediately.
  • Demand refunds for wasted credits and time.
  • Amplify this message across every platform.
  • Expose the gap between Google’s grandiose marketing claims and the pathetic reality users actually experience.

Google has spent years in public through overhyped announcements. It is time the public forces them
This is not just a bad product.
This is one of the most sophisticated consumer rip-offs in modern tech history.
Google Antigravity is not frontier technology.

2

The “engineered exhaustion” theory is genuinely the funniest part of this because it assumes Google actually planned something. Everything about Antigravity screams they didn’t.

What you described is a bug. Token cap hits, the wrapper panics instead of handling it, retries the whole generation, hits the cap again, loops, then eventually forces a compressed output the model never intended to write. You burned 10x your quota and got a worse result. If this is a credit extraction scheme, whoever designed it is terrible at crime. You don’t drain credits by making the product visibly broken. Users just cancel.

The actual question nobody’s asking: why does a hard output token cap even exist inside a coding IDE wrapper? On the raw API it makes sense, you set it yourself per use case. But in a tool designed for long agentic tasks, what problem does it solve? It interrupts completions, breaks multi-step tasks, and apparently triggers a retry spiral that burns more compute than just letting the model finish. It serves no purpose. For anyone.

Just a wrapper team that shipped something broken and then went completely silent. Which is worse honestly.