Skip to content
Live
Trending storyPast story

DeepMind releases Gemini Robotics 2.5 robot model

4 reports4 outlets8 days agoUpdate

The story so far

AI overview

Google DeepMind has released the visual-language-action (VLA) model Gemini Robotics 2.5, claiming it is the most capable VLA model to date, and is gradually being rolled out to robotics partners. DeepMind states that on over 200 tasks (including using tongs to divide meals, laboratory pipetting), the model's success rate has improved by 30% to 100% compared to previous baselines, with speeds 20% to 30% faster; the Apptronik Apollo 2 humanoid robot equipped with this model can now tie a bow with five fingers. In pilot assembly line tests at Mercedes-Benz in Kecskemét, Hungary, and Sindelfingen, Germany, the Apollo 2 completed the sorting and stacking of metal parts using the model, with a reported success rate of 96%. The model supports the Apollo 2 and Franka Duo platforms and adopts real-time control, increasing the success rate of safe task completion by 33% when near obstacles and by 43% when near people. DeepMind has also launched the open-source safety benchmark 'ASIMOV-jailbreak'; IT Home reports that this benchmark has increased the jailbreak resistance rate to 99.5%. The above success rates and claims of near-human-level performance are from DeepMind and reports, not independently verified results.

AI-written from the reports · Updated 8 days ago

Coverage timeline

Follow the coverage to see every side of the story.

11 Nov
  1. ITHome · AI
    Google's strongest VLA robot model, Gemini Robotics 2.5, debuts, can tie bows, already on the Mercedes production line

    Google launches VLA robot model Gemini Robotics 2.5, achieving leaps in dexterity, reliability, and safety, with a success rate of 96% on the Mercedes production line. The model has tested 200 tasks, with a success rate 30%-100% higher than the baseline, a speed increase of 20%-30%, and is compatible with robots like Apptronik Apollo 2, adopting ASIMOV-Jailbreak benchmark to enhance jailbreak resistance to 99.5%.

10 Nov
  1. Ars Technica · AI
    DeepMind reveals Gemini Robotics 2.5 and open safety benchmark "ASIMOV-jailbreak"

    DeepMind releases Gemini Robotics 2.5, whose robotic AI has approached human level in completing 200 tasks such as gripping food and laboratory pipetting, and has achieved high-precision operations like tying bows on the Apptronik Apollo humanoid robot; this version significantly improves dexterity while launching the open-source security benchmark "ASIMOV-jailbreak".

  2. Google DeepMindFeatured
    Gemini Robotics 2.5 accelerates real-world general dexterity

    Google DeepMind released Gemini Robotics 2.5, claiming it is the most capable vision-language-action (VLA) model to date, with a 30% to 100% increase in success rate on 200 tasks compared to previous baselines, and 20% to 30% faster task completion speed.

Heat over time

Now
3
Peak
1412 Nov, 06:00
24h change
-81%

Hover or tap the chart to see each hour; use the arrow keys on a keyboard.