MOJO prediction gives exit status 127 - r

Question

I created a MOJO file with H2O and try to predict based on the saved modell with R. However, I get an exit status 127 error and could not find any solution.
I have updated to the latest H2O release. The issues only occurs with Windows, running the code on Kaggle gives no error at all.
> modelpath <- h2o.download_mojo(fileh, path=getwd(), get_genmodel_jar=TRUE)
> modelpath
[1] "DeepLearning_model_R_1560926535628_2.zip"
> if(is.vector(X)) X <- t(X)
> X <- as.matrix(X)
> Ynew <- h2o.mojo_predict_df(frame = X, mojo_zip_path=modelpath, verbose=T)
...
input_csv: C:\Users\PAPERS~1\AppData\Local\Temp\Rtmpem4zDz/input.csv
mojo_zip:C:\Users\paperspace\Documents\DeepLearning_model_R_1560926535628_2.zip genmodel_jar: C:/Users/paperspace/Documents/h2o-genmodel.jar
output_csv: C:\Users\PAPERS~1\AppData\Local\Temp\Rtmpem4zDz/prediction.csv
classpath: C:/Users/paperspace/Documents/h2o-genmodel.jar
java_options: -Xmx4g -XX:ReservedCodeCacheSize=256m
java cmd: java -Xmx4g -XX:ReservedCodeCacheSize=256m -cp C:/Users/paperspace/Documents/h2o-genmodel.jar hex.genmodel.tools.PredictCsv --mojo C:\Users\paperspace\Documents\DeepLearning_model_R_1560926535628_2.zip --input C:\Users\PAPERS~1\AppData\Local\Temp\Rtmpem4zDz/input.csv --output C:\Users\PAPERS~1\AppData\Local\Temp\Rtmpem4zDz/prediction.csv --decimal
Fehler in h2o.mojo_predict_csv(input_csv_path = input_csv_path, mojo_zip_path = mojo_zip_path, :
SYSTEM COMMAND FAILED (exit status 127)
Zusätzlich: Warnmeldung:
In dir.create(tmp_dir) :
'C:\Users\PAPERS~1\AppData\Local\Temp\Rtmpem4zDz' existiert bereits