lognorm.ppf scale of percentile

Question

If I want to calculate say 5-th percentile of the data fitted using log-normal distribution I use:

lognorm.ppf(0.05, sigma, loc, scale)

I get the answer X

Is X the 5-th percentile of my original data before transformation to log?

Or do I have to transform it using:

X_original = np.exp(X)

I tried asking ChatGPT and every time I get a different answer