Google releases Gemini Robotics 2.5 VLA model and launches pilot on Mercedes-Benz production lines
Understand the story
On November 10, 2026, Google DeepMind released the next-generation vision-language-action (VLA) model Gemini Robotics 2.5, which it calls the strongest VLA model to date. According to first-hand information from DeepMind, in evaluations across more than 200 tasks, Gemini Robotics 2.5 improved task success rates by 30% to 100% over previous baselines, and boosted task completion speed by 20% to 30%. The same day, Ars Technica reported that alongside releasing the robotics platform, Google DeepMind also open-sourced a safety benchmark called ASIMOV-Jailbreak. On November 11, The Decoder reported on the model's progress in industrial settings: on active Mercedes-Benz production lines in Kecskemét, Hungary, and Sindelfingen, Germany, Apollo 2 humanoid robots used Gemini Robotics 2.5 to complete kit-sorting tasks with a 96% success rate. The reports so far have not disclosed further details about the model's architecture or commercialization plans.
AI overview of reporting · Updated 12 min前
Reporting timeline
Explore different aspects of the story through its reporting.
- The DecoderGoogle DeepMind's Gemini Robotics 2.5 scores 96 percent sorting metal parts in a Mercedes-Benz assembly pilot
Google DeepMind released the vision-language-action model Gemini Robotics 2.5, piloted on active Mercedes-Benz production lines in Kecskemét, Hungary and Sindelfingen, Germany, where Apollo 2 humanoid robots completed kit sorting with a 96% success rate.
- Ars Technica · AIDeepMind reveals Gemini Robotics 2.5 and open safety benchmark "ASIMOV-jailbreak"
Google DeepMind has released its next-generation robotics platform Gemini Robotics 2.5, which improves task success rates by 30 to 100% and is 20 to 30% faster than the previous version.
- Google DeepMindSelectedGemini Robotics 2.5 accelerates real-world general dexterity
Google DeepMind has released its strongest vision-language-action (VLA) model yet, Gemini Robotics 2.5. Evaluation across more than 200 tasks shows success rates improved by 30% to 100% over previous baselines, with task completion speed up 20% to 30%.
Story heat trend
Current heat14·Comparable peak15（11/11 00:00）·Comparable change over 24 hours –
The trend compares the same continuously observed sources, which may cover fewer sources than the current heat score. Hover or select the chart to inspect each hour; use the left and right arrow keys to navigate.