Google Unveils Gemini Robotics 2.5 with Enhanced Dexterity
Overview
Google DeepMind launched Gemini Robotics 2.5, a VLA model claiming up to 100% higher success rates on tasks and a 96% rate in Mercedes-Benz validation with Apptronik's Apollo 2 robot.
Generated from attributed reports · Updated 1 days ago
Event evidence and corrections
1 attributed source owners. Ownership does not establish independent confirmation. Quantities are reported separately and are never added together.
Control: reported autonomous
“在此次概念验证中，Apollo 2 自主完成汽车物料拣配工作：这是一项重复性强、体力消耗大的任务，需要将散料箱中的单个零部件，按照线上每台车辆的精确需求顺序，分拣到规整托盘内。”
Exact source · revision 2ithome
Reported quantity · Reported success rate: 96 percent · reported trials · Differing source assertions
“可靠性 ：在奔驰生产线的真实概念验证中，该模型助力 Apptronik 的 Apollo 2 人形机器人拣选并打包汽车零部件，成功率达 96% ，可承接单调工作，让技术人员专注于复杂装配任务。”
Exact source · revision 2ithome
Reported quantity · Reported success rate: 30 percent · Basis not reported · Differing source assertions
“评估显示，该模型在任务精度、鲁棒性与速度上均有显著提升：相较于此前基线，Gemini Robotics 2.5 按任务类型不同，成功率高出 30% 至 100% ，任务完成速度快 20% 至 30% 。”
Exact source · revision 2ithome
Reported quantity · Reported success rate: 100 percent · Basis not reported · Differing source assertions
“评估显示，该模型在任务精度、鲁棒性与速度上均有显著提升：相较于此前基线，Gemini Robotics 2.5 按任务类型不同，成功率高出 30% 至 100% ，任务完成速度快 20% 至 30% 。”
Exact source · revision 2ithome
Developments
- 2026-11-10 15:00 UTC · 2 reportsGemini Robotics 2.5 ReleaseIT Home — Robotics：Google Unveils Gemini Robotics 2.5, a VLA Model with Enhanced Dexterity and Reliability
- 2026-11-09 19:13 UTC · 1 reportsLive demo of fine-tuned Gemini Robotics model on Apptronik robot sorting partsChris Paxton — X Robotics：Watching appronik robot sort parts - this is a live demo of a fine-tuned gemini robotics model
Report timeline
Follow attributed reports and material updates.
- IT Home — RoboticsSignalGoogle Unveils Gemini Robotics 2.5, a VLA Model with Enhanced Dexterity and Reliability
Google has launched Gemini Robotics 2.5, its most capable visual-language-action (VLA) robot model to date, which significantly expands the breadth of learnable skills and learning speed for robots. The model improves dexterity, reliability, and safety, enabling robots to operate in unstructured real-world environments. Tested across over 200 tasks, it achieves up to 100% higher success rates on certain tasks compared to previous baselines. In a concept validation with Apptronik's Apollo 2 humanoid robot at Mercedes-Benz production lines, the model achieved a 96% success rate in picking and packing automotive parts.
- Google DeepMind — X RoboticsGemini Robotics 2.5 is here. 🤖 Our latest physical AI is able to learn hundreds of everyday skills - from folding laundry and serving pastries to autonomously sorting car parts on the factory floor. Here’s a snapshot of its capabilities 🧵
Gemini Robotics 2.5 is here. 🤖 Our latest physical AI is able to learn hundreds of everyday skills - from folding laundry and serving pastries to autonomously sorting car parts on the factory floor. Here’s a snapshot of its capabilities 🧵
- Chris Paxton — X RoboticsWatching appronik robot sort parts - this is a live demo of a fine-tuned gemini robotics model
Watching appronik robot sort parts - this is a live demo of a fine-tuned gemini robotics model
Event coverage history
Current attention 16·Peak within the comparable range 22（2026-11-10 23:00 UTC）·Change within the comparable range over 24 hours –
The trend compares participants observed continuously; its scope may be narrower than current attention totals. Hover or click to explore hourly observations, or use the left and right arrow keys.
Timezone · UTC
Article dates follow your selected timezone. Briefing editions use Hong Kong time (UTC+8).