AI Trend Notifier
EN
← wiki

$ cat wiki/models/gemini-robotics-2.md

Gemini Robotics 2

Spec

AttributeValue
DeveloperGoogle DeepMind
Released2026-07-28
Announced2026-07-28
Context windowunknown
Pricingunknown
Licenseunknown
Availabilityearly-access partnership only (not public)
Full nameGemini Robotics 2 (vision-language-action model)
ArchitectureVLA — single learned policy over the full humanoid

Release Date

Announced 2026-07-28 as the vision-language-action member of a three-model family. The embodied-reasoning sibling Gemini Robotics ER 2 followed on 2026-07-30 and is the only one of the three available publicly (source).

ModelTypeAvailability
Gemini Robotics 2VLAearly-access partnership
Gemini Robotics ER 2embodied-reasoning VLMpublic (Gemini API, AI Studio)
Gemini Robotics On-Device 2on-device VLAearly-access partnership

Benchmarks

DeepMind describes this as its first AI to control a full humanoid — legs, torso, arms and multi-finger hands — under one learned policy (source).

Whole-body manipulation, on the Apollo 2 humanoid:

TaskSuccess
Picking from a shelf76.3%
Picking from the floor45.7%
Fine-motor dexterity, on the SharpaWave hand (22 degrees of freedom):
TaskSuccess
Unscrewing a light bulb92%
Tying a trash bag44%
Sealing a ziplock bag40%
Screwing a bulb in36%
Dustpan task32%
The spread is the result. Removing a light bulb is near-solved at 92%; putting one back is 36%,
and the deformable-object tasks (trash bag 44%, ziplock 40%) sit near the bottom. Insertion and
deformable manipulation remain unsolved at this release.

Gemini Robotics On-Device 2 is reported to adapt to a completely new robot embodiment — different shape, sensors and degrees of freedom — in a few hours with fewer than 200 demonstration examples.

Use Cases

  • Humanoid whole-body tasks requiring locomotion and manipulation in one policy: walking, crouching and manipulating while reasoning through the task
  • Multi-robot collaboration, orchestrated through Gemini Robotics ER 2
  • Cross-embodiment transfer via the On-Device 2 variant

Compared To

ModelScopeAvailabilityDeveloper
Gemini Robotics 2full humanoid, one policyearly accessGoogle DeepMind
Gemini Robotics ER 1.6embodied reasoning onlyGemini APIGoogle DeepMind
Robostral Navigate8B navigation, single RGB cameraMistral AI
Cosmos-H-Dreamssurgical robotics world modelApache 2.0NVIDIA
No shared benchmark connects this release to Gemini Robotics ER 1.6 — the ER 1.6
headline figure (gauge reading 23% → 93%) has no counterpart here, so the two cannot be ranked
against each other (source).

Safety

Released alongside ASIMOV-Agentic, an open benchmark for agentic robotic safety — see Gemini Robotics ER 2, where the evaluated behaviours sit.

Referenced by

Sources