SUBSCRIBE

Google says its new robot brain is the most capable ever. Google is also the only one keeping score.

Every number in the launch is Google's own. No outside lab has checked the model before it goes near real-world robots.

01THE CLAIM
"Google DeepMind says Gemini Robotics ER 2 is its most capable embodied reasoning model, reporting 91.3% accuracy on timing benchmarks and the highest accuracy across core capabilities." [SOURCE ↗]

THE MOVE: SELF-MARKED, graded by the party that benefits from the grade

TRUE, BUT3 SOURCES · LIVE 2026-09-05
GOOGLE DEEPMIND TRACK RECORD5 CLAIMS · 45/100 BS RATE →
91.3%GOOGLE'S TIMING SCORE
0.96sEXECUTION SPEED
JUL 30LAUNCH
Google says its new robot brain is the most capable ever. Google is also the only one keeping score.
02THE CHECK

THE CLAIM. Gemini Robotics ER 2 is Google DeepMind's most capable embodied reasoning model, with 91.3% timing accuracy and the highest accuracy across every core capability.

THE CHECK. read the fine print on 'highest accuracy'. Highest against what? Every benchmark cited traces back to Google's own evaluations, and the tech press repeats the same 91.3% without an independent run.

THE TWIST. the model may well be excellent. Nobody outside Google has confirmed it, and 'most capable' with a single scorer is a press release, not a result.

03SAY THIS IN THE MEETING
"'Which of these numbers came from someone other than Google?' If the honest answer is none, treat 'most capable' as a slogan."

On July 30, 2026, Google DeepMind launched Gemini Robotics ER 2, calling it "our most capable embodied reasoning" model. The launch reporting carries the numbers as delivered: 91.3% accuracy on timing benchmarks, a 0.96s mean absolute distance execution speed, and the highest accuracy across core ca

🔒 THE FULL AUTOPSY · FREE WITH AN ACCOUNT

You just read the free check. Sign in free, a code by email, no passwords, and the rest unlocks: the evidence trail, the steelman and the rebuttal, all 3 sources with quotes and screenshots, and our on-record call.

This story is a stable, citable object. If you can falsify a verdict, tell us. Corrections are loud here.