Google DeepMind's Gemini Robotics 2.5 hits 96% in Mercedes-Benz parts-sorting pilot
Source The Decoder

At a glance
Google DeepMind has released Gemini Robotics 2.5, an AI model that helps robots see, understand instructions, and move. In a pilot on Mercedes-Benz factory lines, a humanoid robot reportedly sorted metal parts with a 96% success rate.
Key points
The model takes in camera images and language instructions and turns them into robot arm and hand movements.
On real Mercedes-Benz lines in Hungary and Germany, Apptronik's Apollo 2 robot sorted parts into trays in the order each car needed.
The company tested it on more than 200 tasks and reported 30 to 100% higher success than its own earlier models.
Google DeepMind also released ASIMOV-Jailbreak, an open-source test that checks whether robots can be tricked into unsafe actions.
Why it matters
It shows how robots are beginning to take on real factory work, which helps readers understand where the robotics industry is heading.
Original article
Google DeepMind's Gemini Robotics 2.5 scores 96 percent sorting metal parts in a Mercedes-Benz assembly pilot
The Decoder · Nov 11, 2026, 2:11 AM GMT+9 · the-decoder.com
Read on The DecoderAI read the original article and summarized it in plain language. It may contain mistakes, so please check the original for accuracy. The article's copyright belongs to The Decoder.
Also reported by
- Unite.AIApollo 2 휴머노이드, Mercedes-Benz 분류 개념 증명에서 96% 성공률 달성
- 디지털데일리휴머노이드가 벤츠 공장서 부품 분류…구글 딥마인드, 로봇 AI 2.5 버전 공개
- 인공지능신문구글, 휴머노이드 로봇에 ‘제미나이 로보틱스 2.5’ 탑재…벤츠 생산라인서 작업 성공률 96% 달성
- 시큐리티팩트구글, 벤츠 독일·헝가리 공장에 로봇 AI '제미나이 2.5' 투입… 부품 분류 성공률 96% 달성
- 와우테일구글, ‘제미나이 로보틱스 2.5’ 공개…벤츠와 현장 성능 검증
- AI타임스구글, '제미나이 로보틱스 2.5' 공개…휴머노이드 공장 작업 성공률 96%
- AI넷[제미니 로보틱스 2.5, 메르세데스-벤츠 공장서 96% 성공률 달성] 제미니 로보틱스 2.5는 실제 세계의 일반적인 민첩성을 구현하는 데 초점을 맞춰 개발됐다. 정밀한 손동작과 도구 사용, 섬세하거나 유연한 물체 취급