Benchmarks
Explore the 23 benchmarks behind the index and every task they test.
RoboDojo teamSimulation
RoboDojo SimReleased Jul 2026 · 37 models
42 tasks · Staged rubric
42 tasks · Staged rubric
Generalization (standard and randomized scenes), Precision, Long-horizon, Memory, Open-ended (unseen task specs). 42 tasks.
1SSOLVE79.5
2
Q0-Preview74.2
Q0-Preview74.23
Claude Fable 5.264.9
Claude Fable 5.264.9RoboDojo teamReal robots
RoboDojo RealReleased Jul 2026 · 14 models
18 tasks · Staged rubric
18 tasks · Staged rubric
6 tasks on each of 3 robots, 18 in all. Scored per robot.
1
SLA-Preview42.3
SLA-Preview42.32
InternW0-Δ38.9
InternW0-Δ38.93
OpenWAM-α37.6
OpenWAM-α37.6Robocurve, Inc.Real robots
RoboDojo-RCReleased Sep 2026 · 12 models
6 tasks · Staged rubric
6 tasks · Staged rubric
6 RoboDojo tasks adapted, Cap Pen, Pack and Pour Fruit, Sort Objects, Stack Bowls, Stand Up Bottles, Store In Safe.
1
Dream-159.3
Dream-159.32
Claude Fable 5.257.3
Claude Fable 5.257.33
Gemini 4 Carbon54.5
Gemini 4 Carbon54.5Artificial AnalysisReal robots
AA-RoboticsReleased Oct 2026 · 5 models
8 tasks · Pass/fail + partial
8 tasks · Pass/fail + partial
8 tasks, press a button, place a toy, pour a drink, tidy blocks, match shapes, write hello world, build a tower and measure a box.
1G6GPT-6.2 Astra79.0
2G6GPT-6.2 Sol78.0
3CFClaude Fable 5.275.0
RoboChallengeReal robots
RoboChallengeReleased Oct 2025 · 14 models
60 tasks · Staged rubric
60 tasks · Staged rubric
30 household tabletop tasks, grouped here by robot. Progress score and success rate.
1
Era076.3
Era076.32
DM072.3
DM072.33
GigaBrain-0.168.3
GigaBrain-0.168.3RoboArena teamReal robots
RoboArenaReleased Jun 2025 · 6 models
2,752 tasks · Pairwise preference
2,752 tasks · Pairwise preference
Open-ended tasks chosen by evaluators, mostly pick and place. Ranked by pairwise preference.
1SVSpirit v1.61788
2
DreamZero1735
DreamZero17353AApricot1661
NVIDIA Seattle Robotics LabSimulation
RoboLab-120Released Apr 2026 · 13 models
120 tasks · Code check
120 tasks · Code check
120 tasks. Difficulty (simple, moderate, complex) and skill (visual, relational, procedural), plus reworded instructions.
1AAMP186.0
2
SenseNova-RoboRSI57.1
SenseNova-RoboRSI57.13LLakeside51.0
Ai2Simulation
MolmoSpacesReleased Feb 2026 · 8 models
19 tasks · Code check
19 tasks · Code check
Pick, pick and place, open, close, randomized camera and commonsense tasks.
1MQMagicBrain + Qwen3.7-Plus94.4
2MIMolmoBot-Img92.8
3
MolmoBot92.0
MolmoBot92.0RoboTwin teamSimulation
RoboTwin 2.0Released Jun 2025 · 22 models
50 tasks · Code check
50 tasks · Code check
50 two-arm tasks, each in a clean and a randomized setting.
1CWCloudRobo-WM89.0
2A0Awomo-0.686.5
3
GO-3 Nano Preview85.0
GO-3 Nano Preview85.0RoboCasa teamSimulation
RoboCasa365Released Mar 2026 · 12 models
50 tasks · Code check
50 tasks · Code check
Atomic tasks (seen), multi-step tasks (seen) and multi-step tasks (unseen).
1HHexaShell85.6
2
Awomo-0.5I64.4
Awomo-0.5I64.43Mminionx62.6
Positronic RoboticsReal robots
PhAILReleased Mar 2026 · 7 models
410 tasks · Generic ladder
410 tasks · Generic ladder
Two benchmarks. 286 everyday DROID tasks, and bin-to-bin pick and place of four household objects timed against a human teleoperator.
1
Cosmos3-Nano-Policy68.1
Cosmos3-Nano-Policy68.12
π0.562.4
π0.562.43
GR00T N1.762.3
GR00T N1.762.3Poke & Wiggle GmbHReal robots
Poke & WiggleReleased Sep 2026 · 11 models
10 tasks · Human rating
10 tasks · Human rating
10 tasks in basic skills, high precision, two-arm handling and tool use, from sorting screws to plugging a DC jack and opening a toolbox with a screwdriver. Each on training, unseen in-between and unseen outside placements.
1
Claude Fable 5.265.0
Claude Fable 5.265.02
Gemini 4 Carbon45.0
Gemini 4 Carbon45.03
Claude Opus 5.539.0
Claude Opus 5.539.0Robocurve, Inc.Real robots
DataCenterBenchReleased Oct 2026 · 9 models
10 tasks · Staged rubric
10 tasks · Staged rubric
10 single steps from data-center manuals, controls (knob, key, switch, cover, slider, keypad) and connectors (plug, unplug, Ethernet, cable clip).
1
GPT-6 Astra40.0
GPT-6 Astra40.02
GPT-6.1 Sol38.3
GPT-6.1 Sol38.33
Gemini 4 Argon35.0
Gemini 4 Argon35.0ArmnetReal robots
ArmnetBenchReleased Jul 2026 · 7 models
12 tasks · Operator label
12 tasks · Operator label
12 tasks, 8 one-arm and 4 two-arm, from block stacking to cable clips and towel folding.
1
π0.547.6
π0.547.62
π035.1
π035.13
GR00T N1.729.4
GR00T N1.729.4Intelligent Robotics and Vision LabReal robots
VLA-ReplicaReleased May 2026 · 7 models
18 tasks · Pass/fail
18 tasks · Pass/fail
10 tasks in pick and place, object interaction and counting, plus an out-of-distribution track.
1
π0.553.0
π0.553.02
GR00T N1.742.0
GR00T N1.742.03
MolmoAct239.0
MolmoAct239.0RoboColosseum teamReal robots
RoboColosseum OpenReleased Nov 2026 · 8 models
7,137 tasks · Progress + preference
7,137 tasks · Progress + preference
Fresh instructions written on the spot at each site, in offices, homes, parks and factories.
1
GPT-6 Astra1925
GPT-6 Astra19252
MolmoAct21869
MolmoAct218693
GalaxeaVLA G0.51744
GalaxeaVLA G0.51744RoboColosseum teamReal robots
RoboColosseum FTReleased Nov 2026 · 5 models
20 tasks · Staged rubric
20 tasks · Staged rubric
20 tasks, five per robot, from stacking and drawers to towel folding and loading a washing machine.
1
π0.573.1
π0.573.12
MolmoAct260.5
MolmoAct260.53
GR00T N1.755.0
GR00T N1.755.0Shanghai AI Laboratory, Intern RoboticsSimulation
EBenchReleased Jun 2026 · 12 models
26 tasks · Code check
26 tasks · Code check
26 task types and 794 instances, ranked on a held-out test split.
1
Motubrain 2.073.5
Motubrain 2.073.52
PICA-WAM61.4
PICA-WAM61.43
OpenWAM-α55.3
OpenWAM-α55.3RoboColiseum consortiumSimulation
RoboColiseumReleased Aug 2026 · 19 models
38 tasks · Code check
38 tasks · Code check
38 tasks in four groups, instruction following, spatial reasoning, robustness and general manipulation.
1MSMXI SimWorld v0 Preview96.6
2HHexaShell96.5
3
FF Omnimind94.5
FF Omnimind94.5LIBERO-Pro TeamSimulation
LIBERO-Pro2Released Oct 2026 · 14 models
339 tasks · Code check
339 tasks · Code check
Standard LIBERO plus static and live perturbations to observation, execution, conditions, robot, language and environment.
1
CloudRobo-WM 1.099.6
CloudRobo-WM 1.099.62LVLingbot-VA98.5
3
Fast-WAM97.8
Fast-WAM97.8LIBERO-Plus authorsSimulation
LPLIBERO-Plus
Released Oct 2025 · 8 models
10,030 tasks · Code check
10,030 tasks · Code check
Seven perturbations, camera, robot start, language, lighting, background, sensor noise and layout.
1OOOpenVLA-OFT+79.6
2
OpenVLA-OFT69.6
OpenVLA-OFT69.63
RIPT-VLA68.4
RIPT-VLA68.4Embodied First Exam teamSimulation
Embodied First ExamReleased Oct 2026 · 10 models
498 tasks · Code check
498 tasks · Code check
Standard, real-time and privileged modes. Coding agents only.
1
Claude Fable 5.249.6
Claude Fable 5.249.62
GPT-6 Astra34.4
GPT-6 Astra34.43
GPT-6.1 Sol32.2
GPT-6.1 Sol32.2Ai2Simulation
Paper-reportedReleased 2021 to 2024 · 13 models
105 tasks · Code check
105 tasks · Code check
LIBERO's four suites (spatial, object, goal, long), SimplerEnv robots, CALVIN multi-step chains.
1
X-VLA98.1
X-VLA98.12
OpenVLA-OFT97.1
OpenVLA-OFT97.13
GR00T N1.797.0
GR00T N1.797.0