Poke &Wiggle

Leaderboard

Model

Claude Fable 5.2

Anthropic

#1

Rank · PAW-GEN-10-Agentic

Model information

Website
Website⁠ (opens in a new tab)
Released
November 2026

PAW-GEN-10-Agentic

Share of episodes rated Ok or Excellent.

Summary

Overall Success
65%
Evaluations
900
NominalEvaluated on the placements the model was trained on.InterpolationEvaluated on unseen placements between the training placements.ExtrapolationEvaluated on unseen placements outside the training placements' convex hull.
EnvironmentNominal · D10Trained on ~10 demonstrations per environment.Interpolation · D10Trained on ~10 demonstrations per environment.Extrapolation · D10Trained on ~10 demonstrations per environment.
Basic skills
Sort screws into bins83%#295% CI 66%–93%100%#195% CI 89%–100%87%#195% CI 70%–95%
Datum-corner alignment80%#295% CI 63%–90%83%#295% CI 66%–93%77%#195% CI 59%–88%
Pour screws into the funnel87%#395% CI 70%–95%90%#395% CI 74%–97%93%#195% CI 79%–98%
High precision
Connect DC jack83%#295% CI 66%–93%77%#295% CI 59%–88%43%#295% CI 27%–61%
Route cable through closed hoops47%#195% CI 30%–64%53%#195% CI 36%–70%53%#195% CI 36%–70%
Put tools in standing toolbox77%#195% CI 59%–88%97%#195% CI 83%–99%30%#195% CI 17%–48%
Bimanual
Place loaded boxes in container (bimanual)20%#195% CI 10%–37%10%#395% CI 3%–26%7%#295% CI 2%–21%
Place spray bottles upright, nozzle forward (bimanual handover)100%#195% CI 89%–100%100%#195% CI 89%–100%87%#195% CI 70%–95%
Tool use
Open the toolbox with a screwdriver70%#195% CI 52%–83%0%#295% CI 0%–11%0%#295% CI 0%–11%
Clamp electrical component87%#195% CI 70%–95%80%#195% CI 63%–90%43%#195% CI 27%–61%
Pooled73%#195% CI 68%–78%69%#195% CI 64%–74%52%#195% CI 46%–58%

Spatial Generalization

Spatial generalization for Claude Fable 5.2: SuccessClaude Fable 5.2: Nominal 73% (95% CI 68%–78%), Interpolation 69% (95% CI 64%–74%), Extrapolation 52% (95% CI 46%–58%).Success rate (%)04080Claude Fable 5.2: 73% (95% CI 68%–78%) on NominalClaude Fable 5.2: 69% (95% CI 64%–74%) on InterpolationClaude Fable 5.2: 52% (95% CI 46%–58%) on ExtrapolationNominalInterpolationExtrapolation

Per Environment (Success)

Model performance by environmentClaude Fable 5.2 overall: Sort screws into bins 90%, Datum-corner alignment 80%, Pour screws into the funnel 90%, Connect DC jack 68%, Route cable through closed hoops 51%, Put tools in standing toolbox 68%, Place loaded boxes in container (bimanual) 12%, Place spray bottles upright, nozzle forward (bimanual handover) 96%, Open the toolbox with a screwdriver 23%, Clamp electrical component 70%.Sort screws into binse01Datum-corner alignmente02Pour screws into the funnele03Connect DC jacke04Route cable through closed hoopse05Put tools in standing toolboxe06Place loaded boxes in container (bimanual)e07Place spray bottles upright, nozzle forward (bimanual handover)e08Open the toolbox with a screwdrivere09Clamp electrical componente10Basic skillsHigh precisionBimanualTool use

Rollouts

Mid-training dataM0
Task-specific dataD10
SpatialNominal

One clip per environment from this entry's evaluation, beside where each evaluation episode placed the environment's objects, coloured by its rating. Use the button beside each environment to open its evaluation dataset in the dataset viewer.

  1. Sort screws into bins
    Success
    83%
    Progress
    98%
  2. Pour screws into the funnel
    Success
    87%
    Progress
    87%
  3. Connect DC jack
    Success
    83%
    Progress
    92%
  4. Route cable through closed hoops
    Success
    47%
    Progress
    70%
  5. Open the toolbox with a screwdriver
    Success
    70%
    Progress
    91%
  6. Held-out benchmark environment
    Datum-corner alignment
    Success
    80%
    Progress
    80%
  7. Held-out benchmark environment
    Put tools in standing toolbox
    Success
    77%
    Progress
    60%
  8. Held-out benchmark environment
    Place loaded boxes in container (bimanual)
    Success
    20%
    Progress
    79%
  9. Held-out benchmark environment
    Place spray bottles upright, nozzle forward (bimanual handover)
    Success
    100%
    Progress
    100%
  10. Held-out benchmark environment
    Clamp electrical component
    Success
    87%
    Progress
    93%

Success by spatial configuration

e01 · Sort screws into bins
Nominal: 83% (95% CI 66%–93%)Interpolation: 100% (95% CI 89%–100%)Extrapolation: 87% (95% CI 70%–95%)NominalInterp.Extrap.
e02 · Datum-corner alignment
Nominal: 80% (95% CI 63%–90%)Interpolation: 83% (95% CI 66%–93%)Extrapolation: 77% (95% CI 59%–88%)NominalInterp.Extrap.
e03 · Pour screws into the funnel
Nominal: 87% (95% CI 70%–95%)Interpolation: 90% (95% CI 74%–97%)Extrapolation: 93% (95% CI 79%–98%)NominalInterp.Extrap.
e04 · Connect DC jack
Nominal: 83% (95% CI 66%–93%)Interpolation: 77% (95% CI 59%–88%)Extrapolation: 43% (95% CI 27%–61%)NominalInterp.Extrap.
e05 · Route cable through closed hoops
Nominal: 47% (95% CI 30%–64%)Interpolation: 53% (95% CI 36%–70%)Extrapolation: 53% (95% CI 36%–70%)NominalInterp.Extrap.
e06 · Put tools in standing toolbox
Nominal: 77% (95% CI 59%–88%)Interpolation: 97% (95% CI 83%–99%)Extrapolation: 30% (95% CI 17%–48%)NominalInterp.Extrap.
e07 · Place loaded boxes in container (bimanual)
Nominal: 20% (95% CI 10%–37%)Interpolation: 10% (95% CI 3%–26%)Extrapolation: 7% (95% CI 2%–21%)NominalInterp.Extrap.
e08 · Place spray bottles upright, nozzle forward (bimanual handover)
Nominal: 100% (95% CI 89%–100%)Interpolation: 100% (95% CI 89%–100%)Extrapolation: 87% (95% CI 70%–95%)NominalInterp.Extrap.
e09 · Open the toolbox with a screwdriver
Nominal: 70% (95% CI 52%–83%)Interpolation: 0% (95% CI 0%–11%)Extrapolation: 0% (95% CI 0%–11%)NominalInterp.Extrap.
e10 · Clamp electrical component
Nominal: 87% (95% CI 70%–95%)Interpolation: 80% (95% CI 63%–90%)Extrapolation: 43% (95% CI 27%–61%)NominalInterp.Extrap.