BenchmarkPAW-GEN-10-Agentic
Environmente10Clamp electrical component
- Episodes
- GPT-6 Astra 30 Claude Fable 5.2 30 Claude Opus 5.5 30 DiT-Flow 30 Gemini 4 Argon 30 GR00T N1.7 30 MolmoAct 2 30 Pi 0.5 30 Rho 30
- Total duration
- GPT-6 Astra 47:44 Claude Fable 5.2 80:36 Claude Opus 5.5 91:26 DiT-Flow 6:21 Gemini 4 Argon 156:44 GR00T N1.7 4:22 MolmoAct 2 8:32 Pi 0.5 5:20 Rho 8:28
- Mean duration
- GPT-6 Astra 95.5 s Claude Fable 5.2 161.2 s Claude Opus 5.5 182.9 s DiT-Flow 12.7 s Gemini 4 Argon 313.5 s GR00T N1.7 8.7 s MolmoAct 2 17.1 s Pi 0.5 10.7 s Rho 16.9 s
- Success
- GPT-6 Astra 7% Claude Fable 5.2 80% Claude Opus 5.5 70% DiT-Flow 0% Gemini 4 Argon 3% GR00T N1.7 0% MolmoAct 2 0% Pi 0.5 0% Rho 0%
- Manipulation density
- GPT-6 Astra Share of time manipulating, five definitionsJoints moving73.3%End-effector moving52.4%Gripper engaged71.8%Contact force > 5 N92.6%Gripper engaged or in contact95.7% Claude Fable 5.2 Share of time manipulating, five definitionsJoints moving85.7%End-effector moving66.5%Gripper engaged97.8%Contact force > 5 N51.9%Gripper engaged or in contact97.8% Claude Opus 5.5 Share of time manipulating, five definitionsJoints moving92.1%End-effector moving79.5%Gripper engaged98.6%Contact force > 5 N80.3%Gripper engaged or in contact98.6% DiT-Flow Share of time manipulating, five definitionsJoints moving98.1%End-effector moving93.4%Gripper engaged8.0%Contact force > 5 N53.0%Gripper engaged or in contact55.0% Gemini 4 Argon Share of time manipulating, five definitionsJoints moving32.7%End-effector moving24.2%Gripper engaged14.9%Contact force > 5 N55.1%Gripper engaged or in contact59.6% GR00T N1.7 Share of time manipulating, five definitionsJoints moving100.0%End-effector moving98.6%Gripper engaged65.6%Contact force > 5 N70.5%Gripper engaged or in contact81.6% MolmoAct 2 Share of time manipulating, five definitionsJoints moving99.5%End-effector moving94.7%Gripper engaged70.3%Contact force > 5 N77.6%Gripper engaged or in contact87.7% Pi 0.5 Share of time manipulating, five definitionsJoints moving99.9%End-effector moving97.6%Gripper engaged71.7%Contact force > 5 N68.6%Gripper engaged or in contact87.5% Rho Share of time manipulating, five definitionsJoints moving99.7%End-effector moving95.6%Gripper engaged77.7%Contact force > 5 N82.2%Gripper engaged or in contact90.5%
- SPARC smoothness
- 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. 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. 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
0:002:04