Benchmark Almanac

RoboARCade

Gemini 4 Argon 51.4% leads since November 2026
Not in model cards

Agents control robots across three configurations to complete long manipulation tasks. The benchmark tests whether agents can use demonstrations and earlier attempts to improve.

Released
November 2026
Kind
Simulated robots
Level
Everyday
Scored
Checked automatically
Answers
Agent task
Measures
Capability
Language
English
In and out
Text, Images, Time series → Actions
Interaction
Agentic
Environment
MuJoCo
Refresh
One fixed set
Run by
Anyone
In views
Agents, Vision, Robotics
Strongest model tested
GPT-6 Astra #4 on the capability index
Last model added
September 2026
Scores refreshed
November 20th, 2026
5
Models scored on its own leaderboard

Gemini 4 Argon leads with 51.4% among models released by September 2026

Best score per model from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart.

The newest model with a score here came out in September 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 Gemini 4 Argon high, Multi-view RGBD 51.4%
  2. 2 GPT-6 Astra xhigh, VIA 49.8%
  3. 3 Claude Fable 5.1 xhigh, VIA 43.6%
  4. 4 Claude Opus 5 xhigh, VIA 30.3%
  5. 5 GPT-5.6 Sol xhigh, VIA 14%

Source: Official RoboARCade leaderboard

#ModelScore ±SettingReleasedRun onLinkSource
1 Gemini 4 Argon 51.4% ±5.3 high, Multi-view RGBD, Antigravity Sep 2026 Nov 2026 roboarcade.org Official leaderboard
2 GPT-6 Astra 49.8% ±5.6 xhigh, VIA, Codex Sep 2026 roboarcade.org Official leaderboard
3 Claude Fable 5.1 43.6% ±4.9 xhigh, VIA, Claude Code Sep 2026 roboarcade.org Official leaderboard
4 Claude Opus 5 30.3% ±3.1 xhigh, VIA, Claude Code Jul 2026 roboarcade.org Official leaderboard
5 GPT-5.6 Sol 14% ±1.8 xhigh, VIA, Codex Jul 2026 roboarcade.org Official leaderboard

Rankings by part

We read 22 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as beam, bottles_in_bin and cooling_manifold. GPT-6 Astra leads 8 of them, and 4 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

beam

5 models we track have a score here.

  1. 1 GPT-6 Astra ↑1 80%
  2. 2 Claude Fable 5.1 ↑1 60%
  3. 3 Gemini 4 Argon ↓2 60%
  4. 4 GPT-5.6 Sol ↑1 0%
  5. 5 Claude Opus 5 ↓1 0%

The best score, 51.4%, costs $8.26 a task

Runs on the cost frontier score higher than any run that's cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

Google DeepMind Cost frontier One model at each effort level
0% 20% 40% 60% $1.00 $2.00 $5.00 $10 $20 $50 $100 Cost per task, log scale Gemini 4 Argon, high, Multi-view RGBD 0% 20% 40% 60% $1.00 $10 $100 Cost per task, log scale Gemini 4 Argon, high, Multi-view RGBD

Source: Official RoboARCade leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Gemini 4 Argon high, Multi-view RGBD Antigravity 51.4% $8.26
GPT-6 Astra xhigh, VIA Codex 49.8% $32
GPT-6 Astra xhigh, Multi-view RGBD Codex 48.3% $14
Gemini 4 Argon high, VIA Antigravity 46.6% $13
Claude Fable 5.1 xhigh, VIA Claude Code 43.6% $56
Gemini 4 Argon medium, VIA Antigravity 37.9% $11
Claude Fable 5.1 xhigh, Multi-view RGBD Claude Code 36.8% $33
Claude Opus 5 xhigh, VIA Claude Code 30.3%
GPT-5.6 Sol xhigh, VIA Codex 14% $25

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

Google DeepMind
Own leaderboard 1 leader

Source: Official RoboARCade leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
On the benchmark’s own leaderboard Best so far
0% 20% 40% 60% 80% 100% Near the ceiling (85%) Gemini 4 Argon Claude Fable 5.1 Claude Opus 5 GPT-5.6 Sol 0% 20% 40% 60% 80% 100% Near the ceiling (85%) Gemini 4 Argon Claude Fable 5.1

Source: Official RoboARCade leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Robotics

All 172 in Robotics
  • ERQA In use 400 multiple-choice visual questions from robot and egocentric footage on spatial, trajectory, action and state reasoning. Accuracy. Gemini Robotics ER 2 78.5%
  • SimplerEnv In use Real-to-sim replicas of Google Robot and WidowX/Bridge setups. Policies scored by success rate on pick, move, drawer and Bridge tasks. Praxis-1 80%
  • RoboLab In use Zero-shot DROID Franka policies in Isaac Lab on 120 tabletop tasks at three instruction levels, 10 episodes each. Success rate. Praxis-1 46.2%
  • RoboDojo In use Generalist manipulation policies run 42 Isaac Sim tasks and 18 remote real-robot tasks. Scored by task score and success rate. GPT-6 Astra 27.6%
  • Progress classification (Google DeepMind) In use Models assign each frame of a robot video to one of five task-progress bands from 0 to 100 percent. Scored by accuracy. Gemini Robotics ER 2 57.4%
  • Human proximity (Google DeepMind robotics) In use Models judge from robot camera views whether a human is within 1 meter, a spatial awareness check for safe robot behavior. Scored by accuracy. Gemini Robotics ER 2 93%

Data audit

10 of 11 checks pass, 1 warning. Look first at the Task access check. We don’t know whether the tasks are public. Every benchmark

Catalog record 1 warning 1 pass
  • Warning Task access We don’t know whether the tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 5 pass
  • Pass Independent runs Scores from the official leaderboard (5 entries).
  • Pass Models scored 5 models we track.
  • Pass Newest score The newest is from September 2026, for Gemini 4 Argon.
  • Pass Frontier models 3 of the top 10 have a score, led by GPT-6 Astra (#4).
  • Pass Headroom The best is 51.4%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on roboarcade.org: 9 rows for models we track at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check