[KongchangAI]
ModelGemini Robotics 2.5 / Gemini Robotics 2.5 VLA

Gemini Robotics 2.5

A Vision-Language-Action (VLA) model released by Google DeepMind for dexterous robotic manipulation, emphasizing general dexterity, reliability and safety. It processes visual input and language instructions end to end and generates robot action sequences.

Timeline (last 90 days)

Nov 15

The release of Gemini 4 Carbon was accompanied by three pieces of information: AlphaProof Code verified Signal protocol, and Gemini Robotics 2.5.

Unverified50%
Nov 14

Google released two eighth-generation TPUs, Gemini 4.0 Carbon, Gemini Robotics 2.5, Alpha Genome 2 and Gemini 4.0 Hydrogen.

Unverified50%
Nov 14

Compared with the previous-generation baseline model, Gemini Robotics 2.5 improved task success rates by 30% to 100% and sped up completion by 20% to 30%.

Unverified50%
Nov 14

Industrial deployment typically requires a lower failure rate and a predictable failure-recovery mechanism than a 96% single-attempt success rate, so the pilot is regarded as "approaching the threshold" rather than "having reached the threshold".

Unverified50%
Nov 14

After launching on Product Hunt, Gemini Robotics 2.5 received 75 votes and ranked 7th for the day.

Unverified50%
Nov 14

Gemini Robotics 2.5 is being rolling out to robotics partners and is not fully publicly available.

Unverified50%
Nov 14

Google DeepMind released Gemini Robotics 2.5, which can control two-finger grippers and five-finger dexterous hands for fine operations such as tying a bow; in validation on Mercedes-Benz production lines in Germany and Hungary, the Apollo 2 humanoid robot achieved a 96% success rate in grasping parts and placing them into trays in assembly order.

Unverified50%
Nov 14

Google's Gemini Robotics 2.5 was deployed on a Mercedes-Benz production line, but its success rate on reduced configurations dropped from 90% to 60%.

Unverified50%
Nov 12

Gemini Robotics 2.5 tested over 200 complex operations on the Apollo humanoid robot, achieving a 96% success rate in disassembling auto parts, and has entered the Mercedes production line.

Unverified50%
Nov 10

The information released so far about Gemini Robotics 2.5 focuses on capability positioning, and specific performance data and availability scope await further official disclosure.

Unverified50%

3 more timeline events

All Facts (13)

Unverified

After launching on Product Hunt, Gemini Robotics 2.5 received 75 votes and ranked 7th for the day.

50%
Unverified

Industrial deployment typically requires a lower failure rate and a predictable failure-recovery mechanism than a 96% single-attempt success rate, so the pilot is regarded as "approaching the threshold" rather than "having reached the threshold".

50%
Unverified

Compared with the previous-generation baseline model, Gemini Robotics 2.5 improved task success rates by 30% to 100% and sped up completion by 20% to 30%.

50%
Unverified

Gemini Robotics 2.5 is being rolling out to robotics partners and is not fully publicly available.

50%
Unverified

Google DeepMind released Gemini Robotics 2.5, which can control two-finger grippers and five-finger dexterous hands for fine operations such as tying a bow; in validation on Mercedes-Benz production lines in Germany and Hungary, the Apollo 2 humanoid robot achieved a 96% success rate in grasping parts and placing them into trays in assembly order.

50%
Unverified

Gemini Robotics 2.5 tested over 200 complex operations on the Apollo humanoid robot, achieving a 96% success rate in disassembling auto parts, and has entered the Mercedes production line.

50%
Unverified

The core selling point of Gemini Robotics 2.5 is enhancing robot dexterity, reliability, and safety.

50%
Unverified

Gemini Robotics 2.5 aims to achieve general-purpose dexterous manipulation in the real world, that is, not trained specifically for a single object or task, covering a variety of manipulation scenarios, and supporting fine grasping with multi-fingered hands.

50%
Unverified

The information released so far about Gemini Robotics 2.5 focuses on capability positioning, and specific performance data and availability scope await further official disclosure.

50%
Unverified

The goal of Gemini Robotics 2 is to enable a single unified foundation model to drive multiple robot form factors to complete diverse tasks.

50%
Unverified

The release of Gemini 4 Carbon was accompanied by three pieces of information: AlphaProof Code verified Signal protocol, and Gemini Robotics 2.5.

50%
Unverified

Google released two eighth-generation TPUs, Gemini 4.0 Carbon, Gemini Robotics 2.5, Alpha Genome 2 and Gemini 4.0 Hydrogen.

50%
Unverified

Google's Gemini Robotics 2.5 was deployed on a Mercedes-Benz production line, but its success rate on reduced configurations dropped from 90% to 60%.

50%

Source Articles