DOOMBENCH.COM
Google 🇺🇸 · Gemini Robotics

Gemini Robotics 2

A vision-language-action model for whole-body humanoid control, dexterous manipulation, and transfer across multiple physical robot embodiments.

DOOM SCORE63.6out of 100model risk profile, not the overall index
0 comments · 0 votesOpen discussion

Public discussion is readable by everyone. Sign in to comment, reply, or vote.

No comments yet. Start the discussion.

Send detailed or private feedback →
CURRENT ASSESSMENT · REVISION 5

Why this model scores 63.6

The model controls full humanoids and bi-arm robots and transfers across embodiments, but demonstrations show uneven success and access is limited to early partners. Physical agency raises autonomy, misuse, and containment concerns, while staged deployment and layered physical and AI safety systems reduce immediate reach and residual control difficulty.

Capability70
Autonomy82
Deployment24
Misuse potential64
Control difficulty65
WATCH AND EXPLORE

Reviewed videos from Artificial Intelligence Videos. These links add context and never change a DoomBench score.

Thumbnail for How Gemini Robotics 2 Coordinates Multiple Robots
2:09

How Gemini Robotics 2 Coordinates Multiple Robots

Google DeepMind demonstrates robots sharing object-handling tasks and coordinating a multi-step workflow with Gemini Robotics 2.

The video directly demonstrates the exact model coordinating multiple robots through a shared multi-step task.
Thumbnail for How Gemini Robotics 2 Coordinates a Robot Team
2:32

How Gemini Robotics 2 Coordinates a Robot Team

Gemini Robotics 2 lets separate robots divide a household task, communicate when help is needed and preserve precise manipulation for the final steps.

The video provides direct visual evidence for the exact model's autonomous team coordination and physical task execution.
Thumbnail for How Gemini Robotics 2 Coordinates Whole-Body Tasks
3:00

How Gemini Robotics 2 Coordinates Whole-Body Tasks

Google DeepMind explains how Gemini Robotics 2 combines whole-body control, dexterous manipulation and independent multi-robot reasoning.

The video explains the exact model's combined whole-body control, dexterity and multi-robot reasoning profile.
Thumbnail for How Gemini Robotics 2 Handles Fine Dexterity
2:19

How Gemini Robotics 2 Handles Fine Dexterity

See Gemini Robotics 2 handle fine physical tasks with robot hands and grippers, from flexible materials and bottle tops to zipped storage bags.

The video directly demonstrates the exact model controlling robot hands and grippers for fine physical tasks.
Thumbnail for How Gemini Robotics 2 Controls a Humanoid Body
2:48

How Gemini Robotics 2 Controls a Humanoid Body

Jie Tan explains how Gemini Robotics 2 links embodied reasoning with whole-body control so a humanoid can balance, navigate and complete tasks.

The video directly explains and demonstrates the exact model's embodied reasoning and whole-body humanoid control.
Thumbnail for How Gemini Robotics 2 Handles Delicate Real-World Tasks
2:17

How Gemini Robotics 2 Handles Delicate Real-World Tasks

Google DeepMind shows how Gemini Robotics 2 controls hands and grippers for zip bags, light bulbs, object placement and knot tying.

The video directly demonstrates the exact model manipulating delicate objects with robot hands and grippers.
MODEL-ATTRIBUTED EVIDENCE

News tied to Gemini Robotics 2

The model score of 63.6 rates this model's risk profile. The overall Doom Index of 66.8 measures the complete temporally weighted evidence record. These values answer different questions.

NET MODEL-ATTRIBUTED INDEX CONTRIBUTION+0.05

Each article's current Doom Index contribution is divided equally among the exact models named on that article. This prevents multi-model evidence from being claimed in full on several model pages. Model risk scores use a bounded temporal offset around their technical profile, but never feed back into the overall index.

AUDIT TRAIL

Model score history

  1. R5
    Doom Score 63.6

    Source-backed model availability audit using the model's existing primary or authoritative catalogue evidence. Exact-version evidence chronology replayed under temporal-monthly-pressure-v4.

    12 Aug 2026
  2. R4
    Doom Score 63.6

    Exact-version evidence chronology replayed after run doombench-hourly-news-20260812-142416 under temporal-monthly-pressure-v4.

    12 Aug 2026
  3. R3
    Doom Score 63.9

    Exact-version evidence chronology replayed after run doombench-hourly-news-20260812-132112 under fixed-sensitivity-v3.

    12 Aug 2026
  4. R2
    Doom Score 65.1

    Full-corpus evidence recalculated after run intensive-backfill-20260811-191225 under bounded-corpus-v2.

    11 Aug 2026
  5. R1
    Doom Score 63.6

    Initial source-backed profile for the exact physical-control release.

    11 Aug 2026

Share this page

DoomBench social sharing card for Gemini Robotics 2.

  1. Gemini Robotics 2 by Google has a DoomBench model risk score of 63.6 out of 100, based on five transparent version-specific dimensions rather than the overall index.

  2. Gemini Robotics 2's highest current DoomBench dimension is autonomy at 82.0 out of 100; the profile publishes every component score and its editorial rationale.

  3. DoomBench links 1 source-backed evidence item to Gemini Robotics 2, while keeping the model's risk profile separate from each item's contribution to the live Doom Index.

    https://www.doombench.com/models/google-deepmind-gemini-robotics-2