BenchmarkPAW-GEN-10-Agentic
Environmente07Place loaded boxes in container (bimanual)
- Episodes
- Claude Fable 5.2 30 Claude Opus 5.5 30 DiT-Flow 30 Gemini 4 Argon 30 GPT-6 Astra 30 GR00T N1.7 30 MolmoAct 2 30 Pi 0.5 30 Rho 30
- Total duration
- Claude Fable 5.2 129:25 Claude Opus 5.5 41:07 DiT-Flow 12:50 Gemini 4 Argon 39:59 GPT-6 Astra 54:48 GR00T N1.7 13:52 MolmoAct 2 11:34 Pi 0.5 10:41 Rho 21:38
- Mean duration
- Claude Fable 5.2 258.8 s Claude Opus 5.5 82.2 s DiT-Flow 25.6 s Gemini 4 Argon 80.0 s GPT-6 Astra 109.6 s GR00T N1.7 27.7 s MolmoAct 2 23.1 s Pi 0.5 21.4 s Rho 43.3 s
- Success
- Claude Fable 5.2 10% Claude Opus 5.5 13% DiT-Flow 0% Gemini 4 Argon 7% GPT-6 Astra 30% GR00T N1.7 17% MolmoAct 2 0% Pi 0.5 0% Rho 0%
- Manipulation density
- Claude Fable 5.2 Share of time manipulating, five definitionsJoints moving69.8%End-effector moving36.4%Gripper engaged77.9%Contact force > 5 N79.3%Gripper engaged or in contact91.3% Claude Opus 5.5 Share of time manipulating, five definitionsJoints moving94.7%End-effector moving73.6%Gripper engaged66.4%Contact force > 5 N80.0%Gripper engaged or in contact80.8% DiT-Flow Share of time manipulating, five definitionsJoints moving98.5%End-effector moving90.6%Gripper engaged31.5%Contact force > 5 N59.1%Gripper engaged or in contact62.1% Gemini 4 Argon Share of time manipulating, five definitionsJoints moving89.0%End-effector moving74.3%Gripper engaged55.4%Contact force > 5 N71.4%Gripper engaged or in contact79.7% GPT-6 Astra Share of time manipulating, five definitionsJoints moving80.7%End-effector moving53.9%Gripper engaged67.6%Contact force > 5 N80.8%Gripper engaged or in contact84.0% GR00T N1.7 Share of time manipulating, five definitionsJoints moving99.9%End-effector moving99.3%Gripper engaged64.2%Contact force > 5 N78.3%Gripper engaged or in contact87.6% MolmoAct 2 Share of time manipulating, five definitionsJoints moving100.0%End-effector moving99.2%Gripper engaged58.6%Contact force > 5 N78.2%Gripper engaged or in contact84.8% Pi 0.5 Share of time manipulating, five definitionsJoints moving99.8%End-effector moving92.6%Gripper engaged63.5%Contact force > 5 N67.3%Gripper engaged or in contact89.3% Rho Share of time manipulating, five definitionsJoints moving100.0%End-effector moving99.8%Gripper engaged65.5%Contact force > 5 N67.7%Gripper engaged or in contact83.5%
- SPARC smoothness
- Claude Fable 5.2 Spectral arc length of the motionHow directly the arm moved, read from the shape of its speed spectrum. Unitless measure ≤ -1; the closer to -1, the smoother.The measure ignores how large a movement is, so a large and a small one compare directly. Shown here as the mean over this environment's measured episodes.Cutoff10 HzAmplitude threshold5%Zero-padding16×Signalend-effector translationBalasubramanian et al., On the analysis of movement smoothness, J Neuroeng Rehabil 2015. Claude Opus 5.5 Spectral arc length of the motionHow directly the arm moved, read from the shape of its speed spectrum. Unitless measure ≤ -1; the closer to -1, the smoother.The measure ignores how large a movement is, so a large and a small one compare directly. Shown here as the mean over this environment's measured episodes.Cutoff10 HzAmplitude threshold5%Zero-padding16×Signalend-effector translationBalasubramanian et al., On the analysis of movement smoothness, J Neuroeng Rehabil 2015. DiT-Flow Spectral arc length of the motionHow directly the arm moved, read from the shape of its speed spectrum. Unitless measure ≤ -1; the closer to -1, the smoother.The measure ignores how large a movement is, so a large and a small one compare directly. Shown here as the mean over this environment's measured episodes.Cutoff10 HzAmplitude threshold5%Zero-padding16×Signalend-effector translationBalasubramanian et al., On the analysis of movement smoothness, J Neuroeng Rehabil 2015. Gemini 4 Argon Spectral arc length of the motionHow directly the arm moved, read from the shape of its speed spectrum. Unitless measure ≤ -1; the closer to -1, the smoother.The measure ignores how large a movement is, so a large and a small one compare directly. Shown here as the mean over this environment's measured episodes.Cutoff10 HzAmplitude threshold5%Zero-padding16×Signalend-effector translationBalasubramanian et al., On the analysis of movement smoothness, J Neuroeng Rehabil 2015. GPT-6 Astra Spectral arc length of the motionHow directly the arm moved, read from the shape of its speed spectrum. Unitless measure ≤ -1; the closer to -1, the smoother.The measure ignores how large a movement is, so a large and a small one compare directly. Shown here as the mean over this environment's measured episodes.Cutoff10 HzAmplitude threshold5%Zero-padding16×Signalend-effector translationBalasubramanian et al., On the analysis of movement smoothness, J Neuroeng Rehabil 2015. GR00T N1.7 Spectral arc length of the motionHow directly the arm moved, read from the shape of its speed spectrum. Unitless measure ≤ -1; the closer to -1, the smoother.The measure ignores how large a movement is, so a large and a small one compare directly. Shown here as the mean over this environment's measured episodes.Cutoff10 HzAmplitude threshold5%Zero-padding16×Signalend-effector translationBalasubramanian et al., On the analysis of movement smoothness, J Neuroeng Rehabil 2015. MolmoAct 2 Spectral arc length of the motionHow directly the arm moved, read from the shape of its speed spectrum. Unitless measure ≤ -1; the closer to -1, the smoother.The measure ignores how large a movement is, so a large and a small one compare directly. Shown here as the mean over this environment's measured episodes.Cutoff10 HzAmplitude threshold5%Zero-padding16×Signalend-effector translationBalasubramanian et al., On the analysis of movement smoothness, J Neuroeng Rehabil 2015. Pi 0.5 Spectral arc length of the motionHow directly the arm moved, read from the shape of its speed spectrum. Unitless measure ≤ -1; the closer to -1, the smoother.The measure ignores how large a movement is, so a large and a small one compare directly. Shown here as the mean over this environment's measured episodes.Cutoff10 HzAmplitude threshold5%Zero-padding16×Signalend-effector translationBalasubramanian et al., On the analysis of movement smoothness, J Neuroeng Rehabil 2015. Rho Spectral arc length of the motionHow directly the arm moved, read from the shape of its speed spectrum. Unitless measure ≤ -1; the closer to -1, the smoother.The measure ignores how large a movement is, so a large and a small one compare directly. Shown here as the mean over this environment's measured episodes.Cutoff10 HzAmplitude threshold5%Zero-padding16×Signalend-effector translationBalasubramanian et al., On the analysis of movement smoothness, J Neuroeng Rehabil 2015.
Episode
Held-out benchmark environment
0:004:36